diff --git a/.gitignore b/.gitignore
index 497c021..a2cc488 100644
--- a/.gitignore
+++ b/.gitignore
@@ -17,3 +17,7 @@ __pycache__/
 
 /doc/_apidoc/
 /build
+
+# Ignore all contents of result and results directories
+/result/
+/results/
\ No newline at end of file
diff --git a/.gitmodules b/.gitmodules
index 5955f18..63ec6f5 100644
--- a/.gitmodules
+++ b/.gitmodules
@@ -1,4 +1,3 @@
-[submodule "src/hyperpod_cli/custom_launcher/launcher/nemo/nemo_framework_launcher"]
-	path = src/hyperpod_cli/custom_launcher/launcher/nemo/nemo_framework_launcher
-	url = https://github.com/NVIDIA/NeMo-Framework-Launcher.git
-	branch = 3d41c31
+[submodule "src/hyperpod_cli/sagemaker_hyperpod_recipes"]
+	path = src/hyperpod_cli/sagemaker_hyperpod_recipes
+	url = https://github.com/aws/sagemaker-hyperpod-recipes.git
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 96b92a5..c4e9997 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,5 +1,11 @@
 # Changelog
 
+## v2.0.0 (2024-12-04)
+
+### Features
+
+- feature: The HyperPod CLI now support ([Hyperpod recipes](https://github.com/aws/sagemaker-hyperpod-recipes.git)). The HyperPod recipes enable customers to get started training and fine-tuning popular publicly-available foundation models like Llama 3.1 405B in minutes. Learn more ([here](https://github.com/aws/sagemaker-hyperpod-recipes.git)).
+
 ## v1.0.0 (2024-09-09)
 
 ### Features
diff --git a/README.md b/README.md
index 85728df..98d76d3 100644
--- a/README.md
+++ b/README.md
@@ -24,7 +24,7 @@ This documentation serves as a reference for the available HyperPod CLI commands
 
 ## Overview
 
-The SageMaker HyperPod CLI is a tool that helps submit training jobs to the Amazon SageMaker HyperPod clusters orchestrated by Amazon EKS. It provides a set of commands for managing the full lifecycle of training jobs, including submitting, describing, listing, and canceling jobs, as well as accessing logs and executing commands within the job's containers. The CLI is designed to abstract away the complexity of working directly with Kubernetes for these core actions of managing jobs on SageMaker HyperPod clusters orchestrated by Amazon EKS.
+The SageMaker HyperPod CLI is a tool that helps submit training jobs to the Amazon SageMaker HyperPod clusters orchestrated by Amazon EKS. It provides a set of commands for managing the full lifecycle of training jobs, including submitting, describing, listing, patching and canceling jobs, as well as accessing logs and executing commands within the job's containers. The CLI is designed to abstract away the complexity of working directly with Kubernetes for these core actions of managing jobs on SageMaker HyperPod clusters orchestrated by Amazon EKS.
 
 ## Prerequisites
 
@@ -76,6 +76,10 @@ SageMaker HyperPod CLI currently supports start training job with:
       ```
       hyperpod get-clusters
       ```
+    - Get your HyperPod clusters to show their capacities and quota allocation info for a team.
+      ```
+      hyperpod get-clusters -n hyperpod-ns-<team-name>
+      ```
     - Connect to one HyperPod cluster and specify a namespace you have access to.
       ```
       hyperpod connect-cluster --cluster-name <cluster-name>
@@ -104,11 +108,12 @@ The HyperPod CLI provides the following commands:
 This command lists the available SageMaker HyperPod clusters and their capacity information.
 
 ```
-hyperpod get-clusters [--region <region>] [--clusters <cluster1,cluster2>] [--orchestrator <eks>] [--output <json|table>]
+hyperpod get-clusters [--region <region>] [--clusters <cluster1,cluster2>] [--namespace <namespace>] [--orchestrator <eks>] [--output <json|table>]
 ```
 
 * `region` (string) - Optional. The region that the SageMaker HyperPod and EKS clusters are located. If not specified, it will be set to the region from the current AWS account credentials.
 * `clusters` (list[string]) - Optional. A list of SageMaker HyperPod cluster names that users want to check the capacity for. This is useful for users who know some of their most commonly used clusters and want to check the capacity status of the clusters in the AWS account.
+* `namespace` (string) - Optional. The namespace that users want to check the quota with. Only the SageMaker managed namespaces are supported.
 * `orchestrator` (enum) - Optional. The orchestrator type for the cluster. Currently, `'eks'` is the only available option.
 * `output` (enum) - Optional. The output format. Available values are `table` and `json`. The default value is `json`.
 
@@ -122,19 +127,19 @@ hyperpod connect-cluster --cluster-name <cluster-name> [--region <region>] [--na
 
 * `cluster-name` (string) - Required. The SageMaker HyperPod cluster name to configure with.
 * `region` (string) - Optional. The region that the SageMaker HyperPod and EKS clusters are located. If not specified, it will be set to the region from the current AWS account credentials.
-* `namespace` (string) - Optional. The namespace that you want to connect to. If not specified, this command uses the [Kubernetes namespace](https://kubernetes.io/docs/concepts/overview/working-with-objects/namespaces/) of the Amazon EKS cluster associated with the SageMaker HyperPod cluster in your AWS account.
+* `namespace` (string) - Optional. The namespace that you want to connect to. If not specified, Hyperpod cli commands will auto discover the accessible namespace.
 
 ### Submitting a Job
 
 This command submits a new training job to the connected SageMaker HyperPod cluster.
 
 ```
-hyperpod start-job --job-name <job-name> [--namespace <namespace>] [--job-kind <kubeflow/PyTorchJob>] [--image <image>] [--command <command>] [--entry-script <script>] [--script-args <arg1 arg2>] [--environment <key=value>] [--pull-policy <Always|IfNotPresent|Never>] [--instance-type <instance-type>] [--node-count <count>] [--tasks-per-node <count>] [--label-selector <key=value>] [--deep-health-check-passed-nodes-only] [--scheduler-type <Kueue>] [--queue-name <queue-name>] [--priority <priority>] [--auto-resume] [--max-retry <count>] [--restart-policy <Always|OnFailure|Never|ExitCode>] [--volumes <volume1,volume2>] [--persistent-volume-claims <claim1:/mount/path,claim2:/mount/path>] [--results-dir <dir>] [--service-account-name <account>]
+hyperpod start-job --job-name <job-name> [--namespace <namespace>] [--job-kind <kubeflow/PyTorchJob>] [--image <image>] [--command <command>] [--entry-script <script>] [--script-args <arg1 arg2>] [--environment <key=value>] [--pull-policy <Always|IfNotPresent|Never>] [--instance-type <instance-type>] [--node-count <count>] [--tasks-per-node <count>] [--label-selector <key=value>] [--deep-health-check-passed-nodes-only] [--scheduler-type <Kueue SageMaker None>] [--queue-name <queue-name>] [--priority <priority>] [--auto-resume] [--max-retry <count>] [--restart-policy <Always|OnFailure|Never|ExitCode>] [--volumes <volume1,volume2>] [--persistent-volume-claims <claim1:/mount/path,claim2:/mount/path>] [--results-dir <dir>] [--service-account-name <account>]
 ```
 
 * `job-name` (string) - Required. The name of the job.
 * `job-kind` (string) - Optional. The training job kind. The job type currently supported is `kubeflow/PyTorchJob`.
-* `namespace` (string) - Optional. The namespace to use. If not specified, this command uses the [Kubernetes namespace](https://kubernetes.io/docs/concepts/overview/working-with-objects/namespaces/) of the Amazon EKS cluster associated with the SageMaker HyperPod cluster in your AWS account.
+* `namespace` (string) - Optional. The namespace to use. If not specified, this command will first use the namespace when connecting the cluster. Otherwise if namespace is not configured when connecting to the cluster, a namespace that is managed by SageMaker will be auto discovered.
 * `image` (string) - Required. The image used when creating the training job.
 * `pull-policy` (enum) - Optional. The policy to pull the container image. Valid values are `Always`, `IfNotPresent`, and `Never`, as available from the PyTorchJob. The default is `Always`.
 * `command` (string) - Optional. The command to run the entrypoint script. Currently, only `torchrun` is supported.
@@ -146,7 +151,7 @@ hyperpod start-job --job-name <job-name> [--namespace <namespace>] [--job-kind <
 * `tasks-per-node` (int) - Optional. The number of devices to use per instance.
 * `label-selector` (dict[string, list[string]]) - Optional. A dictionary of labels and their values that will override the predefined node selection rules based on the SageMaker HyperPod `node-health-status` label and values. If users provide this field, the CLI will launch the job with this customized label selection.
 * `deep-health-check-passed-nodes-only` (bool) - Optional. If set to `true`, the job will be launched only on nodes that have the `deep-health-check-status` label with the value `passed`.
-* `scheduler-type` (enum) - Optional. The scheduler type to use. Currently, only `Kueue` is supported.
+* `scheduler-type` (enum) - Optional. The scheduler type to use which can be `SageMaker`, `Kueue` or `None`. Default value is `SageMaker`.
 * `queue-name` (string) - Optional. The name of the queue to submit the job to, which is created by the cluster admin users in your AWS account.
 * `priority` (string) - Optional. The priority for the job, which needs to be created by the cluster admin users and match the name in the cluster.
 * `auto-resume` (bool) - Optional. The flag to enable HyperPod resilience job auto resume. If set to `true`, the job will automatically resume after pod or node failure. To enable `auto-resume`, you also should set `restart-policy` to `OnFailure`.
@@ -167,7 +172,7 @@ hyperpod get-job --job-name <job-name> [--namespace <namespace>] [--verbose]
 ```
 
 * `job-name` (string) - Required. The name of the job.
-* `namespace` (string) - Optional. The namespace to describe the job in. If not provided, the CLI will try to describe the job in the namespace set by the user while connecting to the cluster. If provided, and the user has access to the namespace, the CLI will describe the job from the specified namespace.
+* `namespace` (string) - Optional. The namespace to use. If not specified, this command will first use the namespace when connecting the cluster. Otherwise if namespace is not configured when connecting to the cluster, a namespace that is managed by SageMaker will be auto discovered.
 * `verbose` (flag) - Optional. If set to `True`, the command enables verbose mode and prints out more detailed output with additional fields.
 
 ### Listing Jobs
@@ -178,7 +183,7 @@ This command lists all the training jobs in the connected SageMaker HyperPod clu
 hyperpod list-jobs [--namespace <namespace>] [--all-namespaces] [--selector <key=value>]
 ```
 
-* `namespace` (string) - Optional. The namespace to list the jobs in. If not provided, this command lists the jobs in the namespace specified during connecting to the cluster. If the namespace is provided and if the user has access to the namespace, this command lists the jobs from the specified namespace.
+* `namespace` (string) - Optional. The namespace to use. If not specified, this command will first use the namespace when connecting the cluster. Otherwise if namespace is not configured when connecting to the cluster, a namespace that is managed by SageMaker will be auto discovered.
 * `all-namespaces` (flag) - Optional. If set, this command lists jobs from all namespaces the data scientist users have access to. The namespace in the current AWS account credentials will be ignored, even if specified with the `--namespace` option.
 * `selector` (string) - Optional. A label selector to filter the listed jobs. The selector supports the '=', '==', and '!=' operators (e.g., `-l key1=value1,key2=value2`).
 
@@ -191,7 +196,7 @@ hyperpod cancel-job --job-name <job-name> [--namespace <namespace>]
 ```
 
 * `job-name` (string) - Required. The name of the job to cancel.
-* `namespace` (string) - Optional. The namespace to cancel the job in. If not provided, the CLI will try to cancel the job in the namespace set by the user while connecting to the cluster. If provided, and the user has access to the namespace, the CLI will cancel the job from the specified namespace.
+* `namespace` (string) - Optional. The namespace to use. If not specified, this command will first use the namespace when connecting the cluster. Otherwise if namespace is not configured when connecting to the cluster, a namespace that is managed by SageMaker will be auto discovered.
 
 ### Listing Pods
 
@@ -202,7 +207,7 @@ hyperpod list-pods --job-name <job-name> [--namespace <namespace>]
 ```
 
 * `job-name` (string) - Required. The name of the job to list pods for.
-* `namespace` (string) - Optional. The namespace to list the pods in. If not provided, the CLI will list the pods in the namespace set by the user while connecting to the cluster. If provided, and the user has access to the namespace, the CLI will list the pods from the specified namespace.
+* `namespace` (string) - Optional. The namespace to use. If not specified, this command will first use the namespace when connecting the cluster. Otherwise if namespace is not configured when connecting to the cluster, a namespace that is managed by SageMaker will be auto discovered.
 
 ### Accessing Logs
 
@@ -214,7 +219,7 @@ hyperpod get-log --job-name <job-name> --pod <pod-name> [--namespace <namespace>
 
 * `job-name` (string) - Required. The name of the job to get the log for.
 * `pod` (string) - Required. The name of the pod to get the log from.
-* `namespace` (string) - Optional. The namespace to get the log from. If not provided, the CLI will get the log from the pod in the namespace set by the user while connecting to the cluster. If provided, and the user has access to the namespace, the CLI will get the log from the pod in the specified namespace.
+* `namespace` (string) - Optional. The namespace to use. If not specified, this command will first use the namespace when connecting the cluster. Otherwise if namespace is not configured when connecting to the cluster, a namespace that is managed by SageMaker will be auto discovered.
 
 ### Executing Commands
 
@@ -226,6 +231,21 @@ hyperpod exec --job-name <job-name> [-p <pod-name>] [--all-pods] -- <command>
 
 * `job-name` (string) - Required. The name of the job to execute the command within the container of a pod associated with a training job.
 * `bash-command` (string) - Required. The bash command(s) to run.
-* `namespace` (string) - Optional. The namespace to execute the command in. If not provided, the CLI will try to execute the command in the pod in the namespace set by the user while connecting to the cluster. If provided, and the user has access to the namespace, the CLI will execute the command in the pod from the specified namespace.
+* `namespace` (string) - Optional. The namespace to use. If not specified, this command will first use the namespace when connecting the cluster. Otherwise if namespace is not configured when connecting to the cluster, a namespace that is managed by SageMaker will be auto discovered.
 * `pod` (string) - Optional. The name of the pod to execute the command in. You must provide either `--pod` or `--all-pods`.
-* `all-pods` (flag) - Optional. If set, the command will be executed in all pods associated with the job.
\ No newline at end of file
+* `all-pods` (flag) - Optional. If set, the command will be executed in all pods associated with the job.
+
+### Patch Jobs
+
+This command patches a job with certain operation. Currently only `suspend` and `unsuspend` are supported.
+
+```
+hyperpod patch-job suspend --job-name <job-name> [--namespace <namespace>]
+```
+
+```
+hyperpod patch-job unsuspend --job-name <job-name> [--namespace <namespace>]
+```
+
+* `job-name` (string) - Required. The name of the job to be patched.
+* `namespace` (string) - Optional. The namespace to use. If not specified, this command will first use the namespace when connecting the cluster. Otherwise if namespace is not configured when connecting to the cluster, a namespace that is managed by SageMaker will be auto discovered.
\ No newline at end of file
diff --git a/examples/basic-job-example-config.yaml b/examples/basic-job-example-config.yaml
index 3cea7c4..ce6af82 100644
--- a/examples/basic-job-example-config.yaml
+++ b/examples/basic-job-example-config.yaml
@@ -64,8 +64,10 @@ cluster:
     # Mapping to '--namespace' argument in 'start-job' command.
     namespace: kubeflow
     # custom_labels: Optional. Used to specify the name of the queue, which is created by the cluster admin users.
+    # The priority class label is mapped to '--priority' argument in 'start-job' command if your scheduler type is 'SageMaker'.
     # custom_labels:
     #   kueue.x-k8s.io/queue-name: low-priority-queue2
+    #   kueue.x-k8s.io/priority-class: sample-priority
     custom_labels: null
     # priority_class_name: Optional. The priority for the job, which is created by the cluster admin users.
     # Mapping to '--priority' argument in 'start-job' command.
@@ -96,7 +98,11 @@ cluster:
     # To use SageMaker Hyperpod AutoResume functionality, please set it to OnFailure.
     # Mapping to '--restart-policy' argument in 'start-job' command.
     restartPolicy: OnFailure
-
+    # scheduler_type: Optional. Used to decide which type of scheduler to use. Default value is 'SageMaker' which makes
+    # the job only scheduled on queues created via SageMaker. Another valid value is 'Kueue', with this option, queue name
+    # and namespace has to be manually filled out.
+    # scheduler_type: Kueue
+    scheduler_type: SageMaker
 # base_results_dir: Optional. Location to store the results, checkpoints and logs.
 # Mapping to '--results-dir' argument in 'start-job' command.
 base_results_dir: ./result
diff --git a/helm_chart/HyperPodHelmChart/Chart.yaml b/helm_chart/HyperPodHelmChart/Chart.yaml
index b73333c..f1e1662 100644
--- a/helm_chart/HyperPodHelmChart/Chart.yaml
+++ b/helm_chart/HyperPodHelmChart/Chart.yaml
@@ -42,11 +42,7 @@ dependencies:
   - name: neuron-device-plugin
     version: "0.1.0"
     repository: "file://charts/neuron-device-plugin"
-    condition: neuron-device-plugin.devicePlugin.enabled  
-  - name: kueue  
-    version: "0.1.0"
-    repository: "file://charts/kueue"
-    condition: kueue.enabled  
+    condition: neuron-device-plugin.devicePlugin.enabled
   - name: storage
     version: "0.1.0"
     repository: "file://charts/storage"
@@ -75,3 +71,7 @@ dependencies:
     version: "0.1.0"
     repository: "file://charts/namespaced-role-and-bindings"
     condition: namespaced-role-and-bindings.enabled
+  - name: team-role-and-bindings
+    version: "0.1.0"
+    repository: "file://charts/team-role-and-bindings"
+    condition: team-role-and-bindings.enabled
\ No newline at end of file
diff --git a/helm_chart/HyperPodHelmChart/charts/kueue/Chart.yaml b/helm_chart/HyperPodHelmChart/charts/kueue/Chart.yaml
deleted file mode 100644
index 679138c..0000000
--- a/helm_chart/HyperPodHelmChart/charts/kueue/Chart.yaml
+++ /dev/null
@@ -1,7 +0,0 @@
-apiVersion: v2
-name: kueue
-version: 0.1.0
-appVersion: 1.0
-description: A Helm chart for deploying kueue for scheduling
-keywords:
-  - kueue
\ No newline at end of file
diff --git a/helm_chart/HyperPodHelmChart/charts/kueue/templates/priority-class.yaml b/helm_chart/HyperPodHelmChart/charts/kueue/templates/priority-class.yaml
deleted file mode 100644
index 59a5126..0000000
--- a/helm_chart/HyperPodHelmChart/charts/kueue/templates/priority-class.yaml
+++ /dev/null
@@ -1,10 +0,0 @@
-{{- range .Values.priorityClasses }}
-apiVersion: scheduling.k8s.io/v1
-kind: PriorityClass
-metadata:
-  name: {{ .name }}
-value: {{ .value }}
-preemptionPolicy: {{ .preemptionPolicy | default "PreemptLowerPriority" }}
-globalDefault: {{ .globalDefault | default false }}
-description: "{{ .description }}"
-{{- end }}
\ No newline at end of file
diff --git a/helm_chart/HyperPodHelmChart/charts/kueue/templates/queue.yaml b/helm_chart/HyperPodHelmChart/charts/kueue/templates/queue.yaml
deleted file mode 100644
index bcf1cbe..0000000
--- a/helm_chart/HyperPodHelmChart/charts/kueue/templates/queue.yaml
+++ /dev/null
@@ -1,17 +0,0 @@
-apiVersion: kueue.x-k8s.io/v1beta1
-kind: ClusterQueue
-metadata:
-  name: {{ .Values.clusterQueue.name }}
-spec:
-  namespaceSelector: {}
- 
-{{- range .Values.queues }}
----
-apiVersion: kueue.x-k8s.io/v1beta1
-kind: LocalQueue
-metadata:
-  namespace: {{ .namespace }}
-  name: {{ .name }}
-spec:
-  clusterQueue: {{ $.Values.clusterQueue.name }}
-{{- end }}
\ No newline at end of file
diff --git a/helm_chart/HyperPodHelmChart/charts/kueue/values.yaml b/helm_chart/HyperPodHelmChart/charts/kueue/values.yaml
deleted file mode 100644
index 03a3c07..0000000
--- a/helm_chart/HyperPodHelmChart/charts/kueue/values.yaml
+++ /dev/null
@@ -1,21 +0,0 @@
-priorityClasses:
-  - name: priority3
-    namespace: "kubeflow"
-    value: 1000
-    globalDefault: false
-    description: "High priority class with preemption"
-  - name: priority4
-    namespace: "kubeflow"
-    value: 100
-    globalDefault: false
-    description: "Low priority class without preemption"
-
-clusterQueue:
-  name: "cluster-queue1"
-  namespaceSelector: {}
-
-queues:
-  - name: high-priority-queue2
-    namespace: "kubeflow"
-  - name: low-priority-queue2
-    namespace: "kubeflow"
\ No newline at end of file
diff --git a/helm_chart/HyperPodHelmChart/charts/team-role-and-bindings/Chart.yaml b/helm_chart/HyperPodHelmChart/charts/team-role-and-bindings/Chart.yaml
new file mode 100644
index 0000000..df81257
--- /dev/null
+++ b/helm_chart/HyperPodHelmChart/charts/team-role-and-bindings/Chart.yaml
@@ -0,0 +1,5 @@
+apiVersion: v2
+name: team-role-and-bindings
+version: 0.1.0
+appVersion: 1.0
+description: This chart installs the namespaced and cluster roles and bindings for team members
\ No newline at end of file
diff --git a/helm_chart/HyperPodHelmChart/charts/team-role-and-bindings/templates/team-level-cluster-role.yaml b/helm_chart/HyperPodHelmChart/charts/team-role-and-bindings/templates/team-level-cluster-role.yaml
new file mode 100644
index 0000000..ad66f4c
--- /dev/null
+++ b/helm_chart/HyperPodHelmChart/charts/team-role-and-bindings/templates/team-level-cluster-role.yaml
@@ -0,0 +1,29 @@
+{{- $targe_id := .Values.computeQuotaTarget.targetId | required ".Values.computeQuotaTarget.targetId is required." -}}
+kind: ClusterRole
+apiVersion: rbac.authorization.k8s.io/v1
+metadata:
+  name: hyperpod-ns-{{ $targe_id }}-cluster-role
+rules:
+  - apiGroups: [ "" ]
+    resources: [ "namespaces" ]
+    verbs: ["list", "get"]
+  - apiGroups: [ "kueue.x-k8s.io" ]
+    resources: [ "clusterqueues" ]
+    resourceNames: ["hyperpod-ns-{{ $targe_id }}-clusterqueue"]
+    verbs: [ "list", "get", "watch" ]
+  - apiGroups: [ "kueue.x-k8s.io" ]
+    resources: [ "resourceflavors", "workloadpriorityclasses" ]
+    verbs: [ "list", "get", "watch" ]
+---
+apiVersion: rbac.authorization.k8s.io/v1
+kind: ClusterRoleBinding
+metadata:
+  name: hyperpod-ns-{{ $targe_id }}-cluster-role-binding
+subjects:
+  - kind: Group
+    name: hyperpod-ns-{{ $targe_id }}-namespace-level-group
+    apiGroup: rbac.authorization.k8s.io
+roleRef:
+  kind: ClusterRole
+  name: hyperpod-ns-{{ $targe_id }}-cluster-role
+  apiGroup: rbac.authorization.k8s.io
\ No newline at end of file
diff --git a/helm_chart/HyperPodHelmChart/charts/team-role-and-bindings/templates/team-level-namespaced-role.yaml b/helm_chart/HyperPodHelmChart/charts/team-role-and-bindings/templates/team-level-namespaced-role.yaml
new file mode 100644
index 0000000..371d0fe
--- /dev/null
+++ b/helm_chart/HyperPodHelmChart/charts/team-role-and-bindings/templates/team-level-namespaced-role.yaml
@@ -0,0 +1,54 @@
+{{- $targe_id := .Values.computeQuotaTarget.targetId | required ".Values.computeQuotaTarget.targetId is required." -}}
+kind: Role
+apiVersion: rbac.authorization.k8s.io/v1
+metadata:
+  namespace: hyperpod-ns-{{ $targe_id }}
+  name: hyperpod-ns-{{ $targe_id }}-namespace-role
+rules:
+  - apiGroups: [""]
+    resources: ["namespaces"]
+    verbs: ["get", "describe"]
+  - apiGroups: ["batch"]
+    resources: ["jobs"]
+    verbs: ["get", "list", "watch", "create", "update", "patch", "delete"]
+  - apiGroups: [""]
+    resources: ["pods"]
+    verbs: ["create", "get", "watch", "list"]
+  - apiGroups: [""]
+    resources: ["nodes", "nodes/status", "pods/status"]
+    verbs: ["get", "list", "watch"]
+  - apiGroups: [""]
+    resources: ["pods/log"]
+    verbs: ["get", "list"]
+  - apiGroups: ["kubeflow.org"]
+    resources: ["pytorchjobs", "pytorchjobs/status"]
+    verbs: ["get", "list", "watch", "create", "delete", "update"]
+  - apiGroups: [""]
+    resources: ["serviceaccounts"]
+    verbs: ["get", "list"]
+  - apiGroups: [""]
+    resources: ["configmaps"]
+    verbs: ["create", "update", "get", "delete"]
+  - apiGroups: [""]
+    resources: ["secrets"]
+    verbs: ["create", "get", "list", "delete"]
+  - apiGroups: [ "apiextensions.k8s.io" ]
+    resources: [ "customresourcedefinitions" ]
+    verbs: ["list", "get"]
+  - apiGroups: [ "kueue.x-k8s.io" ]
+    resources: [ "workloads", "workloads/status" ]
+    verbs: ["get", "list", "watch", "patch"]
+---
+apiVersion: rbac.authorization.k8s.io/v1
+kind: RoleBinding
+metadata:
+  namespace: hyperpod-ns-{{ $targe_id }}
+  name: hyperpod-ns-{{ $targe_id }}-namespace-role-binding
+subjects:
+- kind: Group
+  name: hyperpod-ns-{{ $targe_id }}-namespace-level-group
+  apiGroup: rbac.authorization.k8s.io
+roleRef:
+  kind: Role
+  name: hyperpod-ns-{{ $targe_id }}-namespace-role
+  apiGroup: rbac.authorization.k8s.io
diff --git a/helm_chart/HyperPodHelmChart/charts/team-role-and-bindings/values.yaml b/helm_chart/HyperPodHelmChart/charts/team-role-and-bindings/values.yaml
new file mode 100644
index 0000000..d5300e6
--- /dev/null
+++ b/helm_chart/HyperPodHelmChart/charts/team-role-and-bindings/values.yaml
@@ -0,0 +1,2 @@
+computeQuotaTarget: 
+  targetId: null
\ No newline at end of file
diff --git a/helm_chart/HyperPodHelmChart/values.yaml b/helm_chart/HyperPodHelmChart/values.yaml
index 463a7fa..ca28a64 100644
--- a/helm_chart/HyperPodHelmChart/values.yaml
+++ b/helm_chart/HyperPodHelmChart/values.yaml
@@ -116,9 +116,6 @@ mlflow:
 trainingOperators:
   enabled: true
 
-kueue:
-  enabled: false
-
 storage:
   enabled: false
 
@@ -128,6 +125,9 @@ cluster-role-and-bindings:
 namespaced-role-and-bindings:
   enable: false
 
+team-role-and-bindings:
+  enabled: false
+
 nvidia-device-plugin:
   devicePlugin:
     enabled: true
diff --git a/helm_chart/install_dependencies.sh b/helm_chart/install_dependencies.sh
index 428d913..ec7b5b1 100755
--- a/helm_chart/install_dependencies.sh
+++ b/helm_chart/install_dependencies.sh
@@ -8,8 +8,6 @@ check_status() {
   fi
 }
 
-echo "Installing kueue......"
-kubectl apply --server-side -f https://github.com/kubernetes-sigs/kueue/releases/download/v0.8.0/manifests.yaml
 check_status
 
 # Add additional dependencies below as needed
diff --git a/helm_chart/readme.md b/helm_chart/readme.md
index 894494a..c55808f 100644
--- a/helm_chart/readme.md
+++ b/helm_chart/readme.md
@@ -16,10 +16,10 @@ chmod 700 get_helm.sh
 | Chart Name                   | Usage                                                                                                                                                                                   | Enable by default |
 |------------------------------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-------------------|
 | Cluster role and binding     | Defines cluster-wide roles and bindings for Kubernetes resources, allowing cluster administrators to assign and manage permissions across the entire cluster.                           | No                |
+| Team role and binging        | Defines cluster and namespaced roles and bindings, allowing cluster administrators to create scientist roles with sufficient permissions to submit jobs to the accessible teams.        | No                |
 | Deep health check            | Implements advanced health checks for Kubernetes services and pods to ensure deep monitoring of resource status and functionality beyond basic liveness and readiness probes.           | Yes               |
 | Health monitoring agent      | Deploys an agent to continuously monitor the health of Kubernetes applications, providing detailed insights and alerting for potential issues.                                          | Yes               |
 | Job auto restart             | Configures automatic restart policies for Kubernetes jobs, ensuring failed or terminated jobs are restarted based on predefined conditions for high availability.                       | Yes               |
-| Kueue                        | Manages Kubernetes workloads using the Kueue scheduling framework, enabling resource queuing and prioritization for jobs across multiple clusters or queues.                            | No                |
 | MLflow                       | Installs the MLflow platform for managing machine learning experiments, tracking models, and storing model artifacts in a scalable manner within the Kubernetes cluster.                | No                |
 | MPI Operators                | Orchestrates MPI (Message Passing Interface) jobs on Kubernetes, providing an efficient way to manage distributed machine learning or high-performance computing (HPC) workloads.       | Yes               |
 | namespaced-role-and-bindings | Creates roles and role bindings within a specific namespace to manage fine-grained access control for Kubernetes resources in a limited scope.                                          | No                |
@@ -41,7 +41,7 @@ helm lint src/hyperpod_cli/helm_chart/HyperPodHelmChart
 Notes:
 1. To use the resiliency feature in SageMaker HyperPod cluster with Amazon EKS, you need to first install this HelmChart to EKS cluster before creating HyperPod cluster.
 
-2. Kueue, MLflow, and other optional components are disabled by default. If you wish to enable them, you'll need to manually update the values.yaml file in the main chart by setting the corresponding feature flags to true. Please note: Kueue need to be installed separately after installing other dependencies through this helm chart. See below Step Four for more details.
+2. MLflow, and other optional components are disabled by default. If you wish to enable them, you'll need to manually update the values.yaml file in the main chart by setting the corresponding feature flags to true.
 
 3. Below <dependencies> is the default release name. helm install <dependencies> is the example.
 
@@ -87,11 +87,21 @@ Notes:
   helm install dependencies helm_chart/HyperPodHelmChart --namespace kube-system
   ```
 
-### Step Four (only required for installing Kueue):
-* First run `./install_dependencies.sh` script under helm_chart folder. This script installs certain CRDs directly via commands.
-* Then run below command to complete installation.
+## 5. Create Team Role
+
+* To create role for hyperpod cluster users, please set the value for `computeQuotaTarget.targeId` when installing or upgrade the chart. This value is the same as the `targeId` of quota allocation.
+  ```
+  helm install dependencies helm_chart/HyperPodHelmChart --namespace kube-system --set computeQuotaTarget.targetId=<target_id>
+  ```
+### Step Five (only required for changing the health monitoring agent installation on your cluster- version upgrade):
+* This command is required to change the version of the Health Monitoring Agent running on your Hyperpod cluster.
+```
+helm upgrade dependencies helm_chart/HyperPodHelmChart/charts/health-monitoring-agent --namespace kube-system -f helm_chart/HyperPodHelmChart/charts/health-monitoring-agent/values.yaml
+```
+
+* To install the sub-chart separately that only contains roles and role bindings
   ```
-  helm upgrade dependencies src/hyperpod_cli/helm_chart/HyperPodHelmChart --values src/hyperpod_cli/helm_chart/HyperPodHelmChart/charts/kueue/values.yaml --namespace kube-system
+  helm install dependencies helm_chart/HyperPodHelmChart/charts/team-role-and-bindings --set computeQuotaTarget.targetId=<target_id>
   ```
 ### Step Five (only required for changing the health monitoring agent installation on your cluster- version upgrade):
 * This command is required to change the version of the Health Monitoring Agent running on your Hyperpod cluster.
@@ -99,7 +109,7 @@ Notes:
 helm upgrade dependencies helm_chart/HyperPodHelmChart/charts/health-monitoring-agent --namespace kube-system -f helm_chart/HyperPodHelmChart/charts/health-monitoring-agent/values.yaml
 ```
 
-## 5. Notes
+## 6. Notes
 - If you intend to use the Health Monitoring Agent container image from another region, please see below list to find relevant region's URI.
   ```
   IAD 767398015722.dkr.ecr.us-east-1.amazonaws.com/hyperpod-health-monitoring-agent:1.0.318.0_1.0.35.0
@@ -117,7 +127,7 @@ helm upgrade dependencies helm_chart/HyperPodHelmChart/charts/health-monitoring-
   GRU 025066253954.dkr.ecr.sa-east-1.amazonaws.com/hyperpod-health-monitoring-agent:1.0.318.0_1.0.35.0
   ```
 
-## 6. Troubleshooting
+## 7. Troubleshooting
 
 #### ServiceAccount "neuron-device-plugin" in namespace "kube-system" exists and cannot be imported into the current release
 - This means dependencies already installed. You can change the name of the dependencies if you want another dependencies. To see dependencies already existed or not
diff --git a/setup.py b/setup.py
index ec5b4e2..9ba6501 100644
--- a/setup.py
+++ b/setup.py
@@ -71,7 +71,7 @@
 setup(
     data_files=examples + k8s_templates + nemo_framework_launcher,
     name="hyperpod",
-    version="1.0.0",
+    version="2.0.0",
     packages=find_packages(where="src", exclude=("test",)),
     install_requires=[
         "click==8.1.7",
diff --git a/src/hyperpod_cli/cli.py b/src/hyperpod_cli/cli.py
index ab21fed..e458a2f 100644
--- a/src/hyperpod_cli/cli.py
+++ b/src/hyperpod_cli/cli.py
@@ -22,6 +22,7 @@
     get_job,
     list_jobs,
     list_pods,
+    patch_job,
     start_job,
 )
 from hyperpod_cli.commands.pod import (
@@ -40,7 +41,7 @@
   * get-job         Show details of a specific training job submitted on HyperPod cluster.
   * list-jobs       List training job on a HyperPod cluster.
   * cancel-job      Cancel training job on a HyperPod cluster.
-
+  * patch-job       Patch a job with specific operation on a HyperPod cluster.
 Troubleshooting and Debugging Commands:
   * get-log         Get logs for a pod of training job running on HyperPod cluster.
   * list-pods       List all pods associated with a training job on HyperPod cluster.
@@ -72,6 +73,7 @@ def cli():
 cli.add_command(get_job)
 cli.add_command(list_jobs)
 cli.add_command(cancel_job)
+cli.add_command(patch_job)
 cli.add_command(exec)
 cli.add_command(list_pods)
 cli.add_command(get_log)
diff --git a/src/hyperpod_cli/clients/kubernetes_client.py b/src/hyperpod_cli/clients/kubernetes_client.py
index 8d61f5a..e0f0f2f 100644
--- a/src/hyperpod_cli/clients/kubernetes_client.py
+++ b/src/hyperpod_cli/clients/kubernetes_client.py
@@ -22,10 +22,19 @@
 from kubernetes.config import (
     KUBE_CONFIG_DEFAULT_LOCATION,
 )
-
+from kubernetes.client.rest import ApiException
 from hyperpod_cli.constants.command_constants import (
+    SAGEMAKER_MANAGED_QUEUE_LABEL,
     TEMP_KUBE_CONFIG_FILE,
 )
+from hyperpod_cli.constants.exception_constants import RESOURCE_NOT_FOUND_CODE
+from hyperpod_cli.constants.kueue_constants import (
+    CLUSTER_QUEUE_PRIORITY_CLASS_CUSTOM_OBJECT_PLURAL,
+    KUEUE_CUSTOM_OBJECT_GROUP, 
+    KUEUE_CUSTOM_OBJECT_VERSION, 
+    WORKLOAD_CUSTOM_OBJECT_PLURAL,
+    WORKLOAD_PRIORITY_CLASS_CUSTOM_OBJECT_PLURAL,
+)
 from hyperpod_cli.constants.pytorch_constants import (
     PYTORCH_CUSTOM_OBJECT_GROUP,
     PYTORCH_CUSTOM_OBJECT_PLURAL,
@@ -56,7 +65,7 @@ def __new__(cls, is_get_capacity: bool = False) -> "KubernetesClient":
     def set_context(
         self,
         context_name: str,
-        namespace: str,
+        namespace: Optional[str],
     ) -> None:
         """
         Set the current context in the kubeconfig file.
@@ -68,13 +77,16 @@ def set_context(
         with open(KUBE_CONFIG_PATH, "r") as file:
             kubeconfig = yaml.safe_load(file)
 
-        # Check if the context exists in the kubeconfig
-        # and update the context namespace
+        # Check if the context exists in the kubeconfig and update the context namespace
+        # when namespace is specified in command line
         exist = False
         for context in kubeconfig["contexts"]:
             if context["name"] == context_name:
-                context["context"]["namespace"] = namespace
-                logger.debug(f"updated the namespace to {namespace}")
+                if namespace is not None:
+                    context["context"]["namespace"] = namespace
+                    logger.debug(f"updated the namespace to {namespace}")
+                else:
+                    context["context"].pop("namespace", None)
                 exist = True
 
         if not exist:
@@ -117,6 +129,20 @@ def get_apps_v1_api(self) -> client.AppsV1Api:
             )
         return client.AppsV1Api(self._kube_client)
 
+    def get_auth_v1_api(self) -> client.AuthorizationV1Api:
+        """
+        Get the AuthorizationV1Api client.
+
+        Returns:
+            client.AuthorizationV1Api: The AuthorizationV1Api client.
+        """
+
+        if self._kube_client is None:
+            raise RuntimeError(
+                "Kubernetes client is not initialized. Call set_context() first."
+            )
+        return client.AuthorizationV1Api(self._kube_client)
+
     def context_exists(self, context: str) -> bool:
         """
         Check if the specified context exists in the kubeconfig.
@@ -171,7 +197,7 @@ def get_current_context_namespace(
         Returns current user namespace context
         """
         contexts, active_context = config.list_kube_config_contexts()
-        return active_context["context"]["namespace"]
+        return active_context["context"]["namespace"] if "namespace" in active_context["context"] else None
 
     def list_namespaces(self) -> List[str]:
         """
@@ -185,6 +211,31 @@ def list_namespaces(self) -> List[str]:
                 result.append(namespace.metadata.name)
         return result
 
+    def get_sagemaker_managed_namespace(self, namespace: Optional[str]):
+        """
+        Verify if a namespace is sagemaker managed. SageMaker created namespaces have custom label
+        attached, indicating that the namespace is SageMaker managed.
+
+        Args:
+            namespace (Optional[str]): The input namespace that will be checked.
+
+        Returns:
+            Optional[V1Namespace]: V1Namespace response if namespace exists and it is verified being managed by SageMaker. Otherwise return None.
+        """
+        if namespace is None:
+            return None
+        try:
+            response = client.CoreV1Api().read_namespace(name=namespace)
+            labels = response.metadata.labels
+            if labels and SAGEMAKER_MANAGED_QUEUE_LABEL in labels and labels[SAGEMAKER_MANAGED_QUEUE_LABEL] == "true":
+                return response
+        except ApiException as e:
+            if e.status == RESOURCE_NOT_FOUND_CODE:
+                return None
+            else:
+                raise e
+        return None
+
     def list_pods_with_labels(self, namespace: str, label_selector: str):
         return client.CoreV1Api().list_namespaced_pod(
             namespace=namespace,
@@ -262,4 +313,37 @@ def exec_command_on_pod(
             command=bash_command,
         )
 
+    def patch_workload(self, workload_name: str, namespace: str, patch_body: str):
+        return client.CustomObjectsApi().patch_namespaced_custom_object(
+            group=KUEUE_CUSTOM_OBJECT_GROUP,
+            version=KUEUE_CUSTOM_OBJECT_VERSION,
+            namespace=namespace,
+            plural=WORKLOAD_CUSTOM_OBJECT_PLURAL,
+            name=workload_name,
+            body=patch_body
+        )
+
+    def get_workload_by_label(self, label_selector: str, namespace: str):
+        return client.CustomObjectsApi().list_namespaced_custom_object(
+            group=KUEUE_CUSTOM_OBJECT_GROUP,
+            version=KUEUE_CUSTOM_OBJECT_VERSION,
+            namespace=namespace,
+            plural=WORKLOAD_CUSTOM_OBJECT_PLURAL,
+            label_selector=label_selector,
+        )
+    
+    def list_workload_priority_classes(self):
+        return client.CustomObjectsApi().list_cluster_custom_object(
+            group=KUEUE_CUSTOM_OBJECT_GROUP,
+            version=KUEUE_CUSTOM_OBJECT_VERSION,
+            plural=WORKLOAD_PRIORITY_CLASS_CUSTOM_OBJECT_PLURAL,
+        )
+    
+    def get_cluster_queue(self, cluster_queue_name: str):
+        return client.CustomObjectsApi().get_cluster_custom_object(
+            group=KUEUE_CUSTOM_OBJECT_GROUP,
+            version=KUEUE_CUSTOM_OBJECT_VERSION,
+            plural=CLUSTER_QUEUE_PRIORITY_CLASS_CUSTOM_OBJECT_PLURAL,
+            name=cluster_queue_name
+        )
     # Add more methods to access other APIs as needed
diff --git a/src/hyperpod_cli/commands/cluster.py b/src/hyperpod_cli/commands/cluster.py
index ed62031..bab4ad2 100644
--- a/src/hyperpod_cli/commands/cluster.py
+++ b/src/hyperpod_cli/commands/cluster.py
@@ -29,10 +29,16 @@
     KubernetesClient,
 )
 from hyperpod_cli.constants.command_constants import (
+    AVAILABLE_ACCELERATOR_DEVICES_KEY,
     DEEP_HEALTH_CHECK_STATUS_LABEL,
     HP_HEALTH_STATUS_LABEL,
+    HYPERPOD_NAMESPACE_PREFIX,
     INSTANCE_TYPE_LABEL,
+    NVIDIA_GPU_RESOURCE_LIMIT_KEY,
     SAGEMAKER_HYPERPOD_NAME_LABEL,
+    SAGEMAKER_MANAGED_CLUSTER_QUEUE_SUFFIX,
+    SAGEMAKER_QUOTA_ALLOCATION_LABEL,
+    TOTAL_ACCELERATOR_DEVICES_KEY,
     Orchestrator,
     TEMP_KUBE_CONFIG_FILE,
     OutputFormat,
@@ -93,17 +99,26 @@
     is_flag=True,
     help="Enable debug mode",
 )
+@click.option(
+    "--namespace",
+    "-n",
+    type=click.STRING,
+    required=False,
+    multiple=True,
+    help="Optional. The namespace that you want to check the capacity for. Only SageMaker managed namespaces are supported.",
+)
 def get_clusters(
     region: Optional[str],
     orchestrator: Optional[str],
     output: Optional[str],
     clusters: Optional[str],
     debug: bool,
+    namespace: Optional[List]
 ):
     """List SageMaker Hyperpod Clusters with cluster metadata.
 
     Example Usage:
-    1. List clusters with JSON output: hyperpod get-clusters.
+    1. List clusters with JSON output: hyperpod get-clusters -n hyperpod-ns-test-team
 
     Output:
         [
@@ -113,16 +128,22 @@ def get_clusters(
                 "TotalNodes": 2,
                 "AcceleratorDevicesAvailable": 1,
                 "NodeHealthStatus=Schedulable": 2,
-                "DeepHealthCheckStatus=Passed": "N/A"
+                "DeepHealthCheckStatus=Passed": "N/A",
+                "Namespaces": {
+                    "hyperpod-ns-test-team": {
+                        "AvailableAcceleratorDevices": 1,
+                        "TotalAcceleratorDevices": 1
+                    }
+                }
             }
         ]
 
-    2. List clusters with table output: hyperpod get-clusters --output table.
+    2. List clusters with table output: hyperpod get-clusters -n hyperpod-ns-test-team --output table
 
     Output:
-         Cluster                | InstanceType   |   TotalNodes | AcceleratorDevicesAvailable   |   NodeHealthStatus=Schedulable | DeepHealthCheckStatus=Passed
-         -----------------------+----------------+--------------+-------------------------------+--------------------------------+-----------------------------
-         hyperpod-eks-cluster-a | ml.g5.2xlarge  |            2 |                              1|                              2 |                          N/A
+         Cluster                | InstanceType   |   TotalNodes | AcceleratorDevicesAvailable   |   NodeHealthStatus=Schedulable | DeepHealthCheckStatus=Passed | hyperpod-ns-test-teamTotalAcceleratorDevices   | hyperpod-ns-test-teamAvailableAcceleratorDevices
+         -----------------------+----------------+--------------+-------------------------------+--------------------------------+------------------------------+------------------------------------------------+----------------------------------------------------
+         hyperpod-eks-cluster-a | ml.g5.2xlarge  |            2 |                              1|                              2 |                          N/A | 1                                              | 1
     """
     if debug:
         set_logging_level(logger, logging.DEBUG)
@@ -175,6 +196,7 @@ def get_clusters(
             region=region,
             temp_config_file=TEMP_KUBE_CONFIG_FILE,
             cluster_capacities=cluster_capacities,
+            namespace=namespace,
         )
         # cluster_capacities will only be updated when the cluster
         # is a valid Hyperpod EKS cluster. This check avoid
@@ -197,10 +219,16 @@ def get_clusters(
         "NodeHealthStatus=Schedulable",
         "DeepHealthCheckStatus=Passed",
     ]
+
+    if namespace is not None:
+        for ns in namespace:
+            headers.append(ns + TOTAL_ACCELERATOR_DEVICES_KEY)
+            headers.append(ns + AVAILABLE_ACCELERATOR_DEVICES_KEY)
     if output == OutputFormat.TABLE.value:
         print(tabulate(cluster_capacities, headers=headers, tablefmt="presto"))
     elif output == OutputFormat.JSON.value:
         json_list = [dict(zip(headers, value)) for value in cluster_capacities]
+        _restructure_output(json_list, namespace)
         print(json.dumps(json_list, indent=4))
 
 
@@ -213,6 +241,7 @@ def rate_limited_operation(
     region: Optional[str],
     temp_config_file: str,
     cluster_capacities: List[List[str]],
+    namespace: Optional[List[str]],
 ) -> None:
     try:
         eks_cluster_arn = validator.validate_cluster_and_get_eks_arn(
@@ -231,21 +260,91 @@ def rate_limited_operation(
         )
         nodes_info = _aggregate_nodes_info(nodes)
 
+        ns_nominal_quota = {}
+        ns_quota_usage = {}
+
+        for ns in namespace:
+            sm_managed_namespace = k8s_client.get_sagemaker_managed_namespace(ns)
+            if sm_managed_namespace:
+                quota_allocation_id = sm_managed_namespace.metadata.labels[SAGEMAKER_QUOTA_ALLOCATION_LABEL]
+                cluster_queue_name = HYPERPOD_NAMESPACE_PREFIX + quota_allocation_id + SAGEMAKER_MANAGED_CLUSTER_QUEUE_SUFFIX
+                cluster_queue = k8s_client.get_cluster_queue(cluster_queue_name)
+                nominal_quota = _get_cluster_queue_nominal_quota(cluster_queue)
+                quota_usage = _get_cluster_queue_quota_usage(cluster_queue)
+                ns_nominal_quota[ns] = nominal_quota
+                ns_quota_usage[ns] = quota_usage
+            else:
+                ns_nominal_quota[ns] = {}
+                ns_quota_usage[ns] = {}
+
         for instance_type, nodes_summary in nodes_info.items():
-            cluster_capacities.append(
-                [
-                    cluster_name,
-                    instance_type,
-                    nodes_summary["total_nodes"],
-                    nodes_summary["accelerator_devices_available"],
-                    nodes_summary["schedulable"],
-                    nodes_summary["deep_health_check_passed"],
-                ]
-            )
+            capacities = [
+                cluster_name,
+                instance_type,
+                nodes_summary["total_nodes"],
+                nodes_summary["accelerator_devices_available"],
+                nodes_summary["schedulable"],
+                nodes_summary["deep_health_check_passed"],
+            ]
+            for ns in namespace:
+                capacities.append(ns_nominal_quota.get(ns).get(instance_type, {}).get(NVIDIA_GPU_RESOURCE_LIMIT_KEY, "N/A"))
+                capacities.append(_get_available_quota(ns_nominal_quota.get(ns), ns_quota_usage.get(ns), instance_type, NVIDIA_GPU_RESOURCE_LIMIT_KEY))
+            cluster_capacities.append(capacities)
     except Exception as e:
         logger.error(f"Error processing cluster {cluster_name}: {e}, continue...")
 
 
+def _get_cluster_queue_nominal_quota(cluster_queue):
+    nominal_quota = {}
+    resource_groups = cluster_queue.get("spec", {}).get("resourceGroups", [])
+    resource_group = resource_groups[0]
+
+    for flavor in resource_group.get("flavors", []):
+        flavor_name = flavor.get("name", "unknown")
+        resources = flavor.get("resources", [])
+        for resource in resources:
+            resource_name = resource.get("name")
+            quota = resource.get("nominalQuota")
+            if flavor_name not in nominal_quota:
+                nominal_quota[flavor_name] = {}
+            if resource_name == NVIDIA_GPU_RESOURCE_LIMIT_KEY:
+                quota = int(quota)
+            nominal_quota[flavor_name][resource_name] = quota
+    
+    return nominal_quota
+
+
+def _get_cluster_queue_quota_usage(cluster_queue):
+    quota_usage = {}
+    flavor_usage = cluster_queue.get("status", {}).get("flavorsUsage", [])
+
+    for flavor in flavor_usage:
+        flavor_name = flavor.get("name", "unknown")
+        resources = flavor.get("resources", [])
+        for resource in resources:
+            resource_name = resource.get("name")
+            usage = resource.get("total")
+            if flavor_name not in quota_usage:
+                quota_usage[flavor_name] = {}
+            if resource_name == NVIDIA_GPU_RESOURCE_LIMIT_KEY:
+                usage = int(usage)
+            quota_usage[flavor_name][resource_name] = usage
+
+    return quota_usage
+
+
+def _get_available_quota(nominal, usage, flavor, resource_name):
+    nominal_quota = nominal.get(flavor, {}).get(resource_name, None)
+    usage_quota = usage.get(flavor, {}).get(resource_name, None)
+
+    # Calculating available quota only supports numeric values right now.
+    # Some resources need to be further processed by parsing unit like memory, e.g 10Gi
+    if nominal_quota is not None and usage_quota is not None:
+        return int(nominal_quota) - int(usage_quota)
+    
+    return "N/A"
+
+
 def _get_hyperpod_clusters(sm_client: boto3.client) -> List[str]:
     cluster_names: List[str] = []
     response = sm_client.list_clusters()
@@ -257,6 +356,24 @@ def _get_hyperpod_clusters(sm_client: boto3.client) -> List[str]:
     return cluster_names
 
 
+def _restructure_output(summary_list, namespaces):
+    if not namespaces:
+        return
+
+    for node_summary in summary_list:
+        node_summary["Namespaces"] = {}
+        for ns in namespaces:
+            available_accelerators = node_summary[ns + AVAILABLE_ACCELERATOR_DEVICES_KEY]
+            total_accelerators = node_summary[ns + TOTAL_ACCELERATOR_DEVICES_KEY]
+            quota_accelerator_info = {
+                AVAILABLE_ACCELERATOR_DEVICES_KEY: available_accelerators,
+                TOTAL_ACCELERATOR_DEVICES_KEY: total_accelerators,
+            }
+            node_summary["Namespaces"][ns] = quota_accelerator_info
+            node_summary.pop(ns + AVAILABLE_ACCELERATOR_DEVICES_KEY, None)
+            node_summary.pop(ns + TOTAL_ACCELERATOR_DEVICES_KEY, None)
+
+
 def _aggregate_nodes_info(
     nodes: List[client.V1Node],
 ) -> Dict[str, Dict[str, Any]]:
@@ -303,7 +420,7 @@ def _aggregate_nodes_info(
         else:
             if not node.status:
                 continue
-            gpu_allocatable = node.status.allocatable.get("nvidia.com/gpu")
+            gpu_allocatable = node.status.allocatable.get(NVIDIA_GPU_RESOURCE_LIMIT_KEY)
             neuron_allocatable = node.status.allocatable.get(
                 "aws.amazon.com/neurondevice"
             )
@@ -339,8 +456,7 @@ def _aggregate_nodes_info(
     "-n",
     type=click.STRING,
     required=False,
-    help="Optional. The namespace that you want to connect to. If not specified, this command uses the [Kubernetes namespace](https://kubernetes.io/docs/concepts/overview/working-with-objects/namespaces/) of the Amazon EKS cluster associated with the SageMaker HyperPod cluster in your AWS account.",
-    default="default",
+    help="Optional. The namespace that you want to connect to. If not specified, Hyperpod cli commands will auto discover the accessible namespace.",
 )
 @click.option(
     "--debug",
@@ -351,7 +467,7 @@ def connect_cluster(
     cluster_name: str,
     region: Optional[str],
     debug: bool,
-    namespace: str = "default",
+    namespace: str,
 ) -> None:
     """
     Connect to a HyperPod EKS cluster.
diff --git a/src/hyperpod_cli/commands/job.py b/src/hyperpod_cli/commands/job.py
index c47447c..1d867d2 100644
--- a/src/hyperpod_cli/commands/job.py
+++ b/src/hyperpod_cli/commands/job.py
@@ -10,6 +10,7 @@
 # distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
 # ANY KIND, either express or implied. See the License for the specific
 # language governing permissions and limitations under the License.
+from collections import defaultdict
 import boto3
 import datetime
 import logging
@@ -21,7 +22,6 @@
 
 import click
 import yaml
-from hydra import compose, initialize_config_dir
 from hyperpod_cli import utils
 
 from hyperpod_cli.constants.command_constants import (
@@ -30,23 +30,41 @@
     HYPERPOD_AUTO_RESUME_ANNOTATION_KEY,
     HYPERPOD_KUBERNETES_JOB_PREFIX,
     HYPERPOD_MAX_RETRY_ANNOTATION_KEY,
+    HYPERPOD_NAMESPACE_PREFIX,
+    KUEUE_JOB_UID_LABEL_KEY,
     KUEUE_QUEUE_NAME_LABEL_KEY,
+    KUEUE_WORKLOAD_PRIORITY_CLASS_LABEL_KEY,
     NODE_AFFINITY_DICT,
     DEEP_HEALTH_CHECK_PASSED_ONLY_NODE_AFFINITY_DICT,
+    SAGEMAKER_MANAGED_LOCAL_QUEUE_SUFFIX,
+    SAGEMAKER_QUOTA_ALLOCATION_LABEL,
+    JobPatchType,
+    SAGEMAKER_TRAINING_LAUNCHER_DIR,
     PullPolicy,
     RestartPolicy,
     PersistentVolumeClaim,
+    SchedulerType,
     Volume,
 )
 from hyperpod_cli.clients.kubernetes_client import (
     KubernetesClient,
 )
-from hyperpod_cli.custom_launcher.main import (
-    main as customer_launcher,
-)
+from hyperpod_cli.constants.kueue_constants import KUEUE_CUSTOM_OBJECT_GROUP, WORKLOAD_CUSTOM_OBJECT_PLURAL
+from hyperpod_cli.constants.pytorch_constants import PYTORCH_CUSTOM_OBJECT_GROUP, PYTORCH_CUSTOM_OBJECT_PLURAL
+
+
+# Add the private sagemaker training launcher to the sys path
+if 'SAGEMAKER_TRAINING_LAUNCHER_DIR' not in os.environ:
+    os.environ['SAGEMAKER_TRAINING_LAUNCHER_DIR'] = SAGEMAKER_TRAINING_LAUNCHER_DIR
+
+launcher_dir = os.environ['SAGEMAKER_TRAINING_LAUNCHER_DIR']
+if launcher_dir not in sys.path:
+    sys.path.append(launcher_dir)
+
 from hyperpod_cli.service.cancel_training_job import (
     CancelTrainingJob,
 )
+from hyperpod_cli.service.discover_namespaces import DiscoverNamespaces
 from hyperpod_cli.service.get_training_job import (
     GetTrainingJob,
 )
@@ -65,24 +83,29 @@
 )
 from hyperpod_cli.validators.job_validator import (
     JobValidator,
+    validate_yaml_content,
+    verify_and_load_yaml,
+)
+from kubernetes.client import (
+    V1ResourceAttributes
 )
 
 logger = setup_logger(__name__)
 
-
 @click.command()
 @click.option(
     "--job-name",
     type=click.STRING,
-    required=True,
-    help="Required. The name of the job to see the details of.",
+    required=False,
+    help="Optional. The name of the job to see the details of.",
 )
 @click.option(
     "--namespace",
     "-n",
     type=click.STRING,
     required=False,
-    help="Optional. The namespace to describe the job in. If not provided, the CLI will try to describe the job in the namespace set by the user while connecting to the cluster. If provided, and the user has access to the namespace, the CLI will describe the job from the specified namespace.",
+    help="Optional. The namespace to use. If not specified, this command will first use the namespace wh connecting the cluster."
+    "Otherwise if namespace is not configured when connecting to the cluster, a namespace that is managed by SageMaker will be auto discovered.",
 )
 @click.option(
     "--verbose",
@@ -128,7 +151,8 @@ def get_job(
     "-n",
     type=click.STRING,
     required=False,
-    help="Optional. The namespace to list the jobs in. If not provided, this command lists the jobs in the namespace specified during connecting to the cluster. If the namespace is provided and if the user has access to the namespace, this command lists the jobs from the specified namespace.",
+    help="Optional. The namespace to use. If not specified, this command will first use the namespace wh connecting the cluster."
+    "Otherwise if namespace is not configured when connecting to the cluster, a namespace that is managed by SageMaker will be auto discovered.",
 )
 @click.option(
     "--all-namespaces",
@@ -184,7 +208,8 @@ def list_jobs(
     "-n",
     type=click.STRING,
     required=False,
-    help="Optional. The namespace to list the pods in. If not provided, the CLI will list the pods in the namespace set by the user while connecting to the cluster. If provided, and the user has access to the namespace, the CLI will list the pods from the specified namespace.",
+    help="Optional. The namespace to use. If not specified, this command will first use the namespace wh connecting the cluster."
+    "Otherwise if namespace is not configured when connecting to the cluster, a namespace that is managed by SageMaker will be auto discovered.",
 )
 @click.option(
     "--debug",
@@ -226,7 +251,8 @@ def list_pods(
     "-n",
     type=click.STRING,
     required=False,
-    help="Optional. The namespace to cancel the job in. If not provided, the CLI will try to cancel the job in the namespace set by the user while connecting to the cluster. If provided, and the user has access to the namespace, the CLI will cancel the job from the specified namespace.",
+    help="Optional. The namespace to use. If not specified, this command will first use the namespace wh connecting the cluster."
+    "Otherwise if namespace is not configured when connecting to the cluster, a namespace that is managed by SageMaker will be auto discovered.",
 )
 @click.option(
     "--debug",
@@ -269,7 +295,8 @@ def cancel_job(
     "--namespace",
     "-n",
     type=click.STRING,
-    help="Optional. The namespace to use. If not specified, this command uses the [Kubernetes namespace](https://kubernetes.io/docs/concepts/overview/working-with-objects/namespaces/) of the Amazon EKS cluster associated with the SageMaker HyperPod cluster in your AWS account.",
+    help="Optional. The namespace to use. If not specified, this command will first use the namespace wh connecting the cluster."
+    "Otherwise if namespace is not configured when connecting to the cluster, a namespace that is managed by SageMaker will be auto discovered.",
 )
 @click.option(
     "--job-kind",
@@ -337,9 +364,9 @@ def cancel_job(
 )
 @click.option(
     "--scheduler-type",
-    type=click.STRING,
-    default="Kueue",
-    help="Optional. The scheduler type to use. Currently, only `Kueue` is supported.",
+    type=click.Choice(SchedulerType.get_values()),
+    default=SchedulerType.get_default().value,
+    help="Optional. The scheduler type to use which can be `SageMaker`, `Kueue` or `None`. Default value is `SageMaker`.",
 )
 @click.option(
     "--queue-name",
@@ -393,6 +420,85 @@ def cancel_job(
     help="Optional. Add a temp directory for containers to store data in the hosts."
     " <volume_name>:</host/mount/path>:</container/mount/path>,<volume_name>:</host/mount/path1>:</container/mount/path1>",
 )
+@click.option(
+    "--recipe",
+    type=click.STRING,
+    required=False,
+    help = """Optional. Recipe which accelerates distributed training jobs.
+            Current supported recipes are as follows: \n
+            For all the recipes, click [here](https://github.com/aws/sagemaker-hyperpod-recipes.git) to learn more. \n
+training/mixtral/hf_mixtral_8x7b_seq8k_gpu_p5x16_pretrain \n
+training/mixtral/hf_mixtral_8x7b_seq8k_gpu_p5x32_pretrain \n
+training/mixtral/hf_mixtral_8x22b_seq8k_gpu_p5x64_pretrain \n
+training/mixtral/hf_mixtral_8x22b_seq16k_gpu_p5x32_pretrain \n
+training/mixtral/hf_mixtral_8x7b_seq16k_gpu_p5x16_pretrain \n
+training/mixtral/hf_mixtral_8x22b_seq16k_gpu_p5x64_pretrain \n
+training/mixtral/hf_mixtral_8x22b_seq8k_gpu_p5x32_pretrain \n
+training/mixtral/hf_mixtral_8x7b_seq16k_gpu_p5x32_pretrain \n
+training/custom_model/falcon \n
+training/mistral/hf_mistral_7b_seq8k_gpu_p5x16_pretrain \n
+training/mistral/hf_mistral_7b_seq8k_gpu_p5x32_pretrain \n
+training/mistral/hf_mistral_7b_seq16k_gpu_p5x16_pretrain \n
+training/mistral/hf_mistral_7b_seq16k_gpu_p5x32_pretrain \n
+training/llama/hf_llama3_8b_seq8k_trn1x4_pretrain \n
+training/llama/hf_llama3_8b_seq8k_trn1_fine_tuning \n
+training/llama/hf_llama3_70b_seq8k_trn1x16_pretrain \n
+training/llama/hf_llama3_70b_seq16k_gpu_p5x32_pretrain \n
+training/llama/hf_llama3_70b_seq8k_gpu_p5x32_pretrain \n
+training/llama/hf_llama3_8b_seq8k_gpu_p5x16_pretrain \n
+training/llama/hf_llama3_8b_seq16k_gpu_p5x32_pretrain \n
+training/llama/hf_llama3_2_11b_seq8k_gpu_p5x4_pretrain \n
+training/llama/hf_llama3_8b_seq16k_gpu_p5x16_pretrain \n
+training/llama/hf_llama3_8b_seq8k_gpu_p5x32_pretrain \n
+training/llama/hf_llama3_2_90b_seq8k_gpu_p5x32_pretrain \n
+training/llama/hf_llama3_70b_seq8k_gpu_p5x64_pretrain \n
+training/llama/hf_llama3_70b_seq16k_gpu_p5x64_pretrain \n
+training/llama/llama2_7b_nemo \n
+training/llama/megatron_llama3_1_8b_nemo \n
+training/llama/p4_hf_llama3_70b_seq8k_gpu \n
+fine-tuning/llama/p4_hf_llama3_8b_seq8k_gpu_fine_tuning \n
+fine-tuning/llama/p4_hf_llama3_70b_seq8k_gpu_lora \n
+fine-tuning/llama/hf_llama3_405b_seq8k_gpu_lora \n
+fine-tuning/llama/hf_llama3_405b_seq16k_gpu_lora \n
+fine-tuning/llama/hf_llama3_405b_seq16k_gpu_qlora \n
+fine-tuning/llama/hf_llama3_8b_seq8k_gpu_fine_tuning \n
+fine-tuning/llama/hf_llama3_70b_seq8k_gpu_fine_tuning \n
+fine-tuning/llama/hf_llama3_8b_seq16k_gpu_lora \n
+fine-tuning/llama/hf_llama3_70b_seq16k_gpu_lora \n
+fine-tuning/llama/p4_hf_llama3_8b_seq8k_gpu_lora \n
+fine-tuning/llama/hf_llama3_70b_seq8k_gpu_lora \n
+fine-tuning/llama/hf_llama3_405b_seq8k_gpu_qlora \n
+fine-tuning/llama/p4_hf_llama3_70b_seq8k_gpu_fine_tuning \n
+fine-tuning/llama/hf_llama3_405b_seq128k_gpu_qlora \n
+fine-tuning/llama/hf_llama3_8b_seq16k_gpu_fine_tuning \n
+fine-tuning/llama/hf_llama3_8b_seq8k_gpu_lora \n
+fine-tuning/llama/hf_llama3_70b_seq16k_gpu_fine_tuning \n
+            """
+)
+@click.option(
+    "--override-parameters",
+    type=click.STRING,
+    help="""Optional. Override parameters for the recipe, Below are based on Hydra syntax. Format: 'key1=value1 key2=value2 ...'
+    hyperpod start-job --recipe fine-tuning/llama/hf_llama3_8b_seq8192_gpu --override-parameters \
+'{
+  "+cluster.persistent_volume_claims.0.claimName":"fsx-claim",
+  "+cluster.persistent_volume_claims.0.mountPath":"data",
+  "recipes.run.name": "name",
+  "recipes.exp_manager.exp_dir": "/data/llama8b",
+  "recipes.trainer.num_nodes": 1,
+  "recipes.model.num_hidden_layers": 4,
+  "recipes.model.num_attention_heads": 8,
+  "recipes.model.max_context_width": 8192,
+  "recipes.model.max_position_embeddings": 8192,
+  "recipes.model.train_batch_size": 1,
+  "recipes.model.data.use_synthetic_data": true,
+  "instance_type": "p5.48xlarge",
+  "container": "container link",
+  "recipes.model.data.train_dir": "/data/datasets/wikicorpus_llama3_tokenized_8k/",
+  "recipes.model.data.val_dir": "/data/datasets/wikicorpus_llama3_tokenized_8k_val/",
+}'
+    """,
+)
 @click.option(
     "--debug",
     is_flag=True,
@@ -424,13 +530,14 @@ def start_job(
     service_account_name: Optional[str],
     persistent_volume_claims: Optional[str],
     volumes: Optional[str],
+    recipe: Optional[str],
+    override_parameters: Optional[str],
     debug: bool,
 ):
     if debug:
         set_logging_level(logger, logging.DEBUG)
+        os.environ['HYDRA_FULL_ERROR'] = '1'
 
-    # Perform Post-Processing parameter validations
-    # TODO: refactor validate_start_job_args to use Click ctx
     ctx = click.get_current_context()
     validate_only_config_file_argument(ctx)
 
@@ -439,9 +546,11 @@ def start_job(
         logger.error("Cannot start Training job due to AWS credentials issue")
         sys.exit(1)
 
-    if not namespace and config_file is None:
-        k8s_client = KubernetesClient()
-        namespace = k8s_client.get_current_context_namespace()
+    if namespace is None and config_file is None:
+        namespace = _get_auto_fill_namespace_for_create_job()
+    
+    launcher_config_path = None
+    launcher_config_file_name = None
 
     if not validator.validate_start_job_args(
         config_file,
@@ -460,12 +569,16 @@ def start_job(
         max_retry,
         namespace,
         entry_script,
+        recipe,
     ):
         sys.exit(1)
 
-    """Submit job with the provided configuration file or directly with CLI
-    arguments."""
+    """
+    Submit job with the provided configuration file or directly with CLI arguments.
+    """
+
     if job_name is not None:
+        logger.info(f"job_name: {job_name}")
         config = yaml.safe_load(KUBERNETES_PYTORCH_JOB_TEMPLATE)
         try:
             # Update the configuration with provided arguments
@@ -475,7 +588,7 @@ def start_job(
             config["training_cfg"]["run"]["name"] = job_name
             config["training_cfg"]["run"]["nodes"] = node_count
             config["env_vars"] = ENV_VARS_DICT if environment is None else environment
-
+            
             if (
                 not _is_accelerator_instance_type(str(instance_type))
                 and tasks_per_node is None
@@ -553,17 +666,33 @@ def start_job(
             else:
                 config["cluster"]["cluster_config"].pop("annotations")
 
+            custom_labels = {}
             _override_or_remove(
                 config["cluster"]["cluster_config"], "pullPolicy", pull_policy
             )
             _override_or_remove(
                 config["cluster"]["cluster_config"], "restartPolicy", restart_policy
             )
+            if queue_name is None:
+                queue_name = _get_auto_fill_queue_name(namespace, scheduler_type)
+                _override_or_remove(
+                    custom_labels, 
+                    KUEUE_QUEUE_NAME_LABEL_KEY,
+                    queue_name,
+                )
+
+            # When scheduler type is SageMaker, we should use WorkloadPriorityClass instead which
+            # should be passed as a custom label. Thus we create a custom label here and map the
+            # priority to 'None'
+            if scheduler_type == SchedulerType.SAGEMAKER.value and priority is not None:
+                custom_labels[KUEUE_WORKLOAD_PRIORITY_CLASS_LABEL_KEY] = priority
+                priority = None
+
             _override_or_remove(
                 config["cluster"]["cluster_config"],
                 "custom_labels",
-                {KUEUE_QUEUE_NAME_LABEL_KEY: queue_name}
-                if queue_name is not None
+                custom_labels
+                if custom_labels
                 else None,
             )
             _override_or_remove(
@@ -578,57 +707,148 @@ def start_job(
             logger.error(f"Config template has unexpected error: {e}")
             sys.exit(1)
 
-        now = datetime.datetime.now()
-        timestamp = now.strftime("%Y%m%d-%H%M%S")
-        filename = f"{HYPERPOD_KUBERNETES_JOB_PREFIX}-{timestamp}.yaml"
-        with open(
-            os.path.join(GENERATED_LAUNCHER_CONFIG_FILE_PATH, filename), "w"
-        ) as file:
-            yaml.dump(config, file, default_flow_style=False)
-
-        logger.debug(
-            f"Configuration file generated in: {GENERATED_LAUNCHER_CONFIG_FILE_PATH + filename}"
-        )
-        launcher_config_path = GENERATED_LAUNCHER_CONFIG_FILE_PATH
-        launcher_config_file_name = filename
-    else:
-        """Start job with the provided configuration file or directly with CLI
-        arguments"""
+        launcher_config_path, launcher_config_file_name = _generate_launcher_config_file(config)
+    elif config_file is not None:
+        """Start job with the provided configuration file"""
         if os.path.isabs(config_file):
             abs_config_file_path = config_file
         else:
             abs_config_file_path = os.path.abspath(config_file)
-        if not validator.validate_start_job_config_yaml(abs_config_file_path):
+
+        # Load the yaml configuration file provided by user
+        config = verify_and_load_yaml(abs_config_file_path)
+        if config is None:
             sys.exit(1)
-        config_path, config_name = os.path.split(abs_config_file_path)
-        launcher_config_path = str(config_path)
-        launcher_config_file_name = str(config_name)
 
-    logger.debug(
-        f"Starting job with config {launcher_config_path}{launcher_config_file_name}"
-    )
+        cluster_config = config.get("cluster").get("cluster_config")
+        namespace = cluster_config.get("namespace", None)
+        scheduler_type = cluster_config.get("scheduler_type", SchedulerType.get_default().value)
+        custom_labels = cluster_config.get("custom_labels", {})
+        custom_labels = {} if custom_labels is None else custom_labels
+        queue_name = custom_labels.get(KUEUE_QUEUE_NAME_LABEL_KEY, None)
+        # Autofill namespace
+        if namespace is None:
+            namespace = _get_auto_fill_namespace_for_create_job()
+            _override_or_remove(
+                config["cluster"]["cluster_config"], "namespace", namespace
+            )
+        
+        # Validate the content of the yaml file
+        if not validate_yaml_content(config):
+            sys.exit(1)
 
-    # Initialize Hydra and call custom launcher with Hydra config to submit job
-    try:
-        with initialize_config_dir(config_dir=launcher_config_path, version_base="1.2"):
-            cfg = compose(config_name=launcher_config_file_name)
-            with suppress_standard_output_context():
-                customer_launcher(cfg)
-    except Exception as e:
-        logger.error(f"Starting job failed due to: {e}")
-        sys.exit(1)
-    finally:
-        # Remove temporary created Launcher config file for submit via CLI argument case
-        if job_name is not None and config_file is None:
-            file_to_delete = os.path.join(
-                launcher_config_path, launcher_config_file_name
+        # Autofill queue name
+        if queue_name is None:
+            queue_name = _get_auto_fill_queue_name(namespace, scheduler_type)
+            _override_or_remove(
+                custom_labels, 
+                KUEUE_QUEUE_NAME_LABEL_KEY,
+                queue_name,
             )
-            if os.path.exists(file_to_delete):
-                os.remove(file_to_delete)
+        # Re-fill the custom_labels
+        _override_or_remove(
+            config["cluster"]["cluster_config"],
+            "custom_labels",
+            custom_labels
+            if custom_labels
+            else None,
+        )
+        # Remove scheduler type from the config
+        _override_or_remove(config["cluster"]["cluster_config"], "scheduler_type", None)
+        launcher_config_path, launcher_config_file_name = _generate_launcher_config_file(config)
+    
+    start_training_job(
+        recipe=recipe,
+        override_parameters=override_parameters,
+        job_name=job_name,
+        config_file=config_file,
+        launcher_config_path=launcher_config_path,
+        launcher_config_file_name=launcher_config_file_name,
+        pull_policy=pull_policy,
+        restart_policy=restart_policy,
+        namespace=namespace,
+        service_account_name=service_account_name,
+        priority_class_name=queue_name,
+        volumes=volumes,
+        persistent_volume_claims=persistent_volume_claims,
+        auto_resume=auto_resume,
+        label_selector=label_selector,
+        max_retry=max_retry,
+        deep_health_check_passed_nodes_only=deep_health_check_passed_nodes_only,
+    )
+
     console_link = utils.get_cluster_console_url()
     print(json.dumps({"Console URL": console_link}, indent=1, sort_keys=False))
 
 
+@click.command()
+@click.argument("patch_type", nargs=1)
+@click.option(
+    "--job-name",
+    type=click.STRING,
+    required=True,
+    help="Required. The name of the job to be patched.",
+)
+@click.option(
+    "--namespace",
+    "-n",
+    type=click.STRING,
+    help="Optional. The namespace to use. If not specified, this command will first use the namespace wh connecting the cluster."
+    "Otherwise if namespace is not configured when connecting to the cluster, a namespace that is managed by SageMaker will be auto discovered.",
+)
+def patch_job(patch_type: str, job_name: str, namespace: Optional[str]):
+
+    if patch_type not in JobPatchType.get_values():
+        logger.error(f"Unsupported patch type: '{patch_type}'")
+        exit(1)
+    
+    if namespace is None:
+        resource_attributes_template = V1ResourceAttributes(
+            verb="patch",
+            group=KUEUE_CUSTOM_OBJECT_GROUP,
+            resource=WORKLOAD_CUSTOM_OBJECT_PLURAL,
+        )
+        namespace = DiscoverNamespaces().discover_accessible_namespace(resource_attributes_template)
+
+    
+    patch_type_enum = JobPatchType(patch_type)
+    k8s_client = KubernetesClient()
+
+    # Step 1: get the pytorch job definition, UID in metadata is what we need to fetch
+    # the corresponding workload managed by kueue
+    training_job = k8s_client.get_job(job_name, namespace)
+    uid = training_job.get("metadata", defaultdict()).get("uid", None)
+    
+    if uid is None:
+        logger.error("Cannot patch the job because uid cannot be found in metadata.")
+        exit(1)
+    
+    # Step 2: get the workload by apply filtering with uid retrieved in step 1. Only one workload
+    # is expected to be returned, otherwise throw error and exit because which workload should be
+    # patched is uncertain. But this should not ever happen.
+    workload_label_selector = KUEUE_JOB_UID_LABEL_KEY + "=" + uid
+    workloads = k8s_client.get_workload_by_label(workload_label_selector, namespace).get("items", [])
+
+    if len(workloads) == 0:
+        logger.error(f"No workload found for the job to be patched: '{job_name}'")
+        exit(1)
+    if len(workloads) > 1:
+        logger.error(f"Only exact one workload is expected to be found for job: '{job_name}', but found {len(workloads)}")
+        exit(1)
+    workload_name = workloads[0].get('metadata', {}).get('name')
+    # Step 3: Decide the patch body based on the job patch type specified in command
+    patch_body = ""
+    if patch_type_enum == JobPatchType.SUSPEND:
+        patch_body = {"spec": {"active": False}}
+        k8s_client.patch_workload(workload_name, namespace, patch_body)
+        logger.info(f"Job {job_name} is suspended.")
+    elif patch_type_enum == JobPatchType.UNSUSPEND:
+        patch_body = {"spec": {"active": True}}
+        k8s_client.patch_workload(workload_name, namespace, patch_body)
+        logger.info(f"Job {job_name} is unsuspended.")
+    else:
+        logger.info("Found unsupported patch type. No operation is performed.")
+
 def _override_or_remove(
     config: Dict,
     key: str,
@@ -637,7 +857,7 @@ def _override_or_remove(
     if value is not None:
         config[key] = value
     else:
-        config.pop(key)
+        config.pop(key, None)
 
 
 def _is_accelerator_instance_type(
@@ -651,6 +871,39 @@ def _is_accelerator_instance_type(
         return True
     return False
 
+def _generate_launcher_config_file(config):
+    now = datetime.datetime.now()
+    timestamp = now.strftime("%Y%m%d-%H%M%S")
+    filename = f"{HYPERPOD_KUBERNETES_JOB_PREFIX}-{timestamp}.yaml"
+    with open(
+        os.path.join(GENERATED_LAUNCHER_CONFIG_FILE_PATH, filename), "w"
+    ) as file:
+        yaml.dump(config, file, default_flow_style=False)
+
+    logger.debug(
+        f"Configuration file generated in: {GENERATED_LAUNCHER_CONFIG_FILE_PATH + filename}"
+    )
+    return (GENERATED_LAUNCHER_CONFIG_FILE_PATH, filename)
+
+def _get_auto_fill_namespace_for_create_job():
+    namespace = None
+    resource_attributes_template = V1ResourceAttributes(
+        verb="create",
+        group=PYTORCH_CUSTOM_OBJECT_GROUP,
+        resource=PYTORCH_CUSTOM_OBJECT_PLURAL,
+    )
+    namespace = DiscoverNamespaces().discover_accessible_namespace(resource_attributes_template)
+    return namespace
+
+def _get_auto_fill_queue_name(namespace, scheduler_type):
+    k8s_client = KubernetesClient()
+    sm_managed_namespace = k8s_client.get_sagemaker_managed_namespace(namespace)
+    queue_name = None
+    # Provide queue name if not provided and scheduler type is SageMaker
+    if sm_managed_namespace and scheduler_type == SchedulerType.SAGEMAKER.value:
+        quota_allocation_id = sm_managed_namespace.metadata.labels[SAGEMAKER_QUOTA_ALLOCATION_LABEL]
+        queue_name = HYPERPOD_NAMESPACE_PREFIX + quota_allocation_id + SAGEMAKER_MANAGED_LOCAL_QUEUE_SUFFIX
+    return queue_name
 
 class suppress_standard_output_context:
     def __enter__(self):
@@ -683,3 +936,124 @@ def validate_only_config_file_argument(ctx):
         raise click.BadParameter(
             f"Please only provide 'config-file' argument if you want to start job with .yaml file."
         )
+
+
+def execute_command(cmd, env=None):
+    try:
+        result = subprocess.run(cmd, check=True, capture_output=True, text=True, env=env)
+        print(result.stdout)
+    except subprocess.CalledProcessError as e:
+        logger.error(f"Command failed with exit code {e.returncode}")
+        logger.error(e.stderr)
+        sys.exit(1)
+    except Exception as e:
+        logger.error(f"An error occurred: {str(e)}")
+        sys.exit(1)
+
+
+def start_training_job(recipe, override_parameters, job_name, config_file, launcher_config_path=None, launcher_config_file_name=None,
+                      pull_policy=None, restart_policy=None, namespace=None,
+                      service_account_name=None, priority_class_name=None, volumes=None, persistent_volume_claims=None,
+                      auto_resume=None, label_selector=None, max_retry=None, deep_health_check_passed_nodes_only=None):
+    
+    logger.info(f"recipe: {recipe}, override_parameters: {override_parameters}, job_name: {job_name}, config_file: {config_file}, launcher_config_path: {launcher_config_path}, launcher_config_file_name: {launcher_config_file_name}")
+    env = os.environ.copy()
+    env['HYDRA_FULL_ERROR'] = '1'
+
+    if recipe is None:
+        logger.debug(f"Starting job with config {launcher_config_path}{launcher_config_file_name}")
+        cmd = [
+            'python3',
+            f'{SAGEMAKER_TRAINING_LAUNCHER_DIR}/main.py',
+            f'--config-path={launcher_config_path}',
+            f'--config-name={launcher_config_file_name}',
+            f'base_results_dir={os.path.abspath(os.path.join(os.getcwd(), "results"))}',
+            'cluster.cluster_type=k8s',
+        ]
+        execute_command(cmd, env)
+    else:
+        cmd = [
+            'python3',
+            f'{SAGEMAKER_TRAINING_LAUNCHER_DIR}/main.py',
+            f'recipes={recipe}',
+            'cluster_type=k8s',
+            'cluster=k8s',
+            f'base_results_dir={os.path.abspath(os.path.join(os.getcwd(), "results"))}',
+        ]
+
+        # Add pull policy if provided
+        if pull_policy:
+            cmd.append(f'cluster.pullPolicy="{pull_policy}"')
+
+        # Add restart policy if provided
+        if restart_policy:
+            cmd.append(f'cluster.restartPolicy="{restart_policy}"')
+
+        # Add namespace if provided
+        if namespace:
+            cmd.append(f'cluster.namespace="{namespace}"')
+
+        # Add service account name if provided
+        if service_account_name:
+            cmd.append(f'cluster.service_account_name="{service_account_name}"')
+
+        # Add priority class name if provided
+        if priority_class_name:
+            cmd.append(f'cluster.priority_class_name="{priority_class_name}"')
+
+        # Add volumes if provided (expecting format: "volumeName1:hostPath1:mountPath1,volumeName2:hostPath2:mountPath2")
+        if volumes:
+            for idx, volume in enumerate(volumes.split(',')):
+                vol_name, host_path, mount_path = volume.split(':')
+                cmd.append(f'+cluster.volumes.{idx}.volumeName="{vol_name}"')
+                cmd.append(f'+cluster.volumes.{idx}.hostPath="{host_path}"')
+                cmd.append(f'+cluster.volumes.{idx}.mountPath="{mount_path}"')
+
+        # Add persistent volume claims if provided (expecting format: "claimName1:mountPath1,claimName2:mountPath2")
+        if persistent_volume_claims:
+            for idx, pvc in enumerate(persistent_volume_claims.split(',')):
+                claim_name, mount_path = pvc.split(':')
+                cmd.append(f'+cluster.persistent_volume_claims.{idx}.claimName="{claim_name}"')
+                cmd.append(f'+cluster.persistent_volume_claims.{idx}.mountPath="{mount_path}"')
+
+        if label_selector:
+            cmd.append(f'+cluster.label_selector={label_selector}')
+        elif deep_health_check_passed_nodes_only:
+            cmd.append(f'+cluster.label_selector={DEEP_HEALTH_CHECK_PASSED_ONLY_NODE_AFFINITY_DICT}')
+
+        if auto_resume:
+            # Set max_retry default to 1
+            if max_retry is None:
+                max_retry = 1
+            annotations = {
+                HYPERPOD_AUTO_RESUME_ANNOTATION_KEY: auto_resume,
+                HYPERPOD_MAX_RETRY_ANNOTATION_KEY: max_retry,
+            }
+            cmd.append(f'+cluster.annotations="{annotations}"')
+        
+        logger.info(f"override_parameters: {override_parameters}")
+        if override_parameters:
+            try:
+                # Parse the JSON string into a dictionary
+                override_dict = json.loads(override_parameters)
+
+                # Convert the dictionary into key=value pairs
+                for key, value in override_dict.items():
+                    if isinstance(value, str):
+                        # Ensure strings are properly quoted
+                        cmd.append(f'{key}="{value}"')
+                    else:
+                        cmd.append(f'{key}={value}')
+            except json.JSONDecodeError as e:
+                logger.error(f"Invalid JSON format: {e}")
+                sys.exit(1)
+
+        print(f"Final command: {' '.join(cmd)}")
+        execute_command(cmd, env)
+
+    if job_name is not None and config_file is None:
+        file_to_delete = os.path.join(launcher_config_path, launcher_config_file_name)
+        if os.path.exists(file_to_delete):
+            os.remove(file_to_delete)
+
+
diff --git a/src/hyperpod_cli/constants/command_constants.py b/src/hyperpod_cli/constants/command_constants.py
index c7b3d5a..e5f6295 100644
--- a/src/hyperpod_cli/constants/command_constants.py
+++ b/src/hyperpod_cli/constants/command_constants.py
@@ -11,6 +11,7 @@
 # ANY KIND, either express or implied. See the License for the specific
 # language governing permissions and limitations under the License.
 from enum import Enum
+from pathlib import Path
 
 GENERATED_LAUNCHER_CONFIG_FILE_PATH = "/tmp/"
 HYPERPOD_KUBERNETES_JOB_PREFIX = "hyperpod-k8s-job"
@@ -26,6 +27,8 @@
     },
 }
 KUEUE_QUEUE_NAME_LABEL_KEY = "kueue.x-k8s.io/queue-name"
+KUEUE_WORKLOAD_PRIORITY_CLASS_LABEL_KEY = "kueue.x-k8s.io/priority-class"
+KUEUE_JOB_UID_LABEL_KEY = "kueue.x-k8s.io/job-uid"
 HYPERPOD_AUTO_RESUME_ANNOTATION_KEY = "sagemaker.amazonaws.com/enable-job-auto-resume"
 HYPERPOD_MAX_RETRY_ANNOTATION_KEY = "sagemaker.amazonaws.com/job-max-retry-count"
 ENV_VARS_DICT = {"NCCL_DEBUG": "INFO"}
@@ -33,8 +36,16 @@
 HP_HEALTH_STATUS_LABEL = "sagemaker.amazonaws.com/node-health-status"
 INSTANCE_TYPE_LABEL = "node.kubernetes.io/instance-type"
 DEEP_HEALTH_CHECK_STATUS_LABEL = "sagemaker.amazonaws.com/deep-health-check-status"
+SAGEMAKER_MANAGED_QUEUE_LABEL= "sagemaker.amazonaws.com/sagemaker-managed-queue"
+SAGEMAKER_QUOTA_ALLOCATION_LABEL = "sagemaker.amazonaws.com/quota-allocation-id"
 TEMP_KUBE_CONFIG_FILE = "/tmp/kubeconfig"
-
+HYPERPOD_NAMESPACE_PREFIX = "hyperpod-ns-"
+SAGEMAKER_MANAGED_LOCAL_QUEUE_SUFFIX = "-localqueue"
+SAGEMAKER_MANAGED_CLUSTER_QUEUE_SUFFIX = "-clusterqueue"
+SAGEMAKER_TRAINING_LAUNCHER_DIR = str(Path("./src/hyperpod_cli/sagemaker_hyperpod_recipes").resolve())
+NVIDIA_GPU_RESOURCE_LIMIT_KEY = "nvidia.com/gpu"
+AVAILABLE_ACCELERATOR_DEVICES_KEY = "AvailableAcceleratorDevices"
+TOTAL_ACCELERATOR_DEVICES_KEY = "TotalAcceleratorDevices"
 
 class PullPolicy(Enum):
     ALWAYS = "Always"
@@ -67,6 +78,25 @@ def __init__(self, claim_name, mount_path):
         self.mount_path = mount_path
 
 
+class SchedulerType(Enum):
+    KUEUE = "Kueue"
+    SAGEMAKER = "SageMaker"
+    NONE = "None"
+
+    def get_values():
+        return [scheduler_type.value for scheduler_type in SchedulerType]
+
+    def get_default():
+        return SchedulerType.SAGEMAKER
+
+class JobPatchType(Enum):
+    SUSPEND = "suspend"
+    UNSUSPEND = "unsuspend"
+
+    def get_values():
+        return [type.value for type in JobPatchType]
+
+
 class Volume:
     volume_name: str
     host_path: str
diff --git a/src/hyperpod_cli/constants/exception_constants.py b/src/hyperpod_cli/constants/exception_constants.py
new file mode 100644
index 0000000..4eab6de
--- /dev/null
+++ b/src/hyperpod_cli/constants/exception_constants.py
@@ -0,0 +1,13 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+#     http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+RESOURCE_NOT_FOUND_CODE = 404
diff --git a/src/hyperpod_cli/constants/kueue_constants.py b/src/hyperpod_cli/constants/kueue_constants.py
new file mode 100644
index 0000000..2c523a7
--- /dev/null
+++ b/src/hyperpod_cli/constants/kueue_constants.py
@@ -0,0 +1,17 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+#     http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+KUEUE_CUSTOM_OBJECT_GROUP = "kueue.x-k8s.io"
+WORKLOAD_CUSTOM_OBJECT_PLURAL = "workloads"
+KUEUE_CUSTOM_OBJECT_VERSION = "v1beta1"
+WORKLOAD_PRIORITY_CLASS_CUSTOM_OBJECT_PLURAL = "workloadpriorityclasses"
+CLUSTER_QUEUE_PRIORITY_CLASS_CUSTOM_OBJECT_PLURAL = "clusterqueues"
diff --git a/src/hyperpod_cli/custom_launcher/__init__.py b/src/hyperpod_cli/custom_launcher/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/src/hyperpod_cli/custom_launcher/examples/custom_script/config.yaml b/src/hyperpod_cli/custom_launcher/examples/custom_script/config.yaml
deleted file mode 100755
index bf18aa0..0000000
--- a/src/hyperpod_cli/custom_launcher/examples/custom_script/config.yaml
+++ /dev/null
@@ -1,88 +0,0 @@
-defaults:
-  - override hydra/job_logging: stdout
-
-hydra:
-  run:
-    dir: .
-  output_subdir: null
-
-git:
-  repo_url: null
-  branch: null
-  commit: null
-
-training_cfg:
-
-  entry_script: ??? # Path to the entry script of training/fine-tuning. This path should be inside container or relative path in git repo
-  script_args:
-    - "--some_args" : "debug"
-    - "--some_other_args" : 1
-  run:
-    name: test_custom # Current run name
-    nodes: 8 # Number of nodes to use for current training
-    ntasks_per_node: 8 # Number of devices to use per node
-
-cluster:
-  # Example slurm cluster
-  #
-  # cluster_type: slurm
-  # instance_type: p4d.24xlarge
-  # cluster_config:
-  #   exclusive: True
-  #   job_name_prefix: testcustom_slurm_
-  #
-  # Example k8s cluster
-  cluster_type: k8s
-  instance_type: ???
-  cluster_config:
-    namespace: default # the namespace to submit job
-    # create customized labels for the PytorchJob and Pods deployed jobs.
-    # Example:
-    #   custom_labels:
-    #     label-key-1: label-value-1
-    #     label-key-2: label-value-2
-    custom_labels: null
-    # create customized annotations for the jobs.
-    # Example:
-    #   annotations:
-    #     annotation-key-1: annotation-value-1
-    #     annotation-key-2: annotation-value-2
-    annotations: null
-    # add service account to job pods
-    # Example:
-    #  serviceAccountName: service_account
-    service_account_name: null
-    # priorityClassName for Kueue scheduler to decide jobs priority
-    priority_class_name: null
-    # Create k8s NodeAffinity to select nodes to deploy jobs which matches required and preferred labels
-    # Structure:
-    #   label_selector:
-    #     required: <required label key-values pair>
-    #     preferred: <preferred label key-values pair>
-    #     weights: <weights list used by preferred labels to get nodes priority>
-    # Example:
-    #   label_selector:
-    #     required:
-    #       example-label-key:
-    #         - expected-label-value-1
-    #         - expected-label-value-2
-    #     preferred:
-    #       preferred-label-key:
-    #         - preferred-label-value-1
-    #         - preferred-label-value-2
-    #     weights:
-    #       - 100
-    label_selector: null
-    # persistent volume, usually used to mount FSx
-    persistentVolumeClaims: null
-
-    pullPolicy: Always # policy to pull container, can be Always, IfNotPresent and Never
-    restartPolicy: Never # restart policy
-
-base_results_dir: ???  # Location to store the results, checkpoints and logs.
-container_mounts: # List of additional paths to mount to container. They will be mounted to same path.
-  - null
-container: ??? # container to use
-
-env_vars:
-  NCCL_DEBUG: null # Logging level for NCCL. Set to "INFO" for debug information
diff --git a/src/hyperpod_cli/custom_launcher/launcher/__init__.py b/src/hyperpod_cli/custom_launcher/launcher/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/src/hyperpod_cli/custom_launcher/launcher/accelerator_devices.py b/src/hyperpod_cli/custom_launcher/launcher/accelerator_devices.py
deleted file mode 100644
index 15880c3..0000000
--- a/src/hyperpod_cli/custom_launcher/launcher/accelerator_devices.py
+++ /dev/null
@@ -1,61 +0,0 @@
-acceleratorDevices = {
-    "p4d.24xlarge": 8,
-    "p4de.24xlarge": 8,
-    "p5.48xlarge": 8,
-    "trn1.2xlarge": 1,
-    "trn1.32xlarge": 16,
-    "trn1n.32xlarge": 16,
-    "g5.xlarge": 1,
-    "g5.2xlarge": 1,
-    "g5.4xlarge": 1,
-    "g5.8xlarge": 1,
-    "g5.12xlarge": 4,
-    "g5.16xlarge": 1,
-    "g5.24xlarge": 4,
-    "g5.48xlarge": 8,
-}
-
-coresPerAcceleratorDevice = {
-    "p4d.24xlarge": 1,
-    "p4de.24xlarge": 1,
-    "p5.48xlarge": 1,
-    "trn1.2xlarge": 2,
-    "trn1.32xlarge": 2,
-    "trn1n.32xlarge": 2,
-    "g5.xlarge": 1,
-    "g5.2xlarge": 1,
-    "g5.4xlarge": 1,
-    "g5.8xlarge": 1,
-    "g5.12xlarge": 1,
-    "g5.16xlarge": 1,
-    "g5.24xlarge": 1,
-    "g5.48xlarge": 1,
-}
-
-
-def get_num_accelerator_devices(instance_type: str):
-    """
-    Get the number of accelerator devices on an instance type.
-    Accelerator device could be GPU or Trainium chips
-    :param instance_type: AWS EC2 instance type
-    :return: number of accelerator devices for the instance type or None if instance
-    type not in the accelerator devices map
-    """
-    if instance_type not in acceleratorDevices:
-        return None
-
-    return acceleratorDevices[instance_type]
-
-
-def get_num_cores_per_accelerator(instance_type: str):
-    """
-    Get the number of cores per accelerator device on an instance type.
-    Currently, Trainium has 2 cores per device while Nvida has 1 core per device.
-    :param instance_type: AWS EC2 instance type
-    :return: number of cores for the accelerator device or None if instance type
-    not in the map
-    """
-    if instance_type not in coresPerAcceleratorDevice:
-        return None
-
-    return coresPerAcceleratorDevice[instance_type]
diff --git a/src/hyperpod_cli/custom_launcher/launcher/config_validator/__init__.py b/src/hyperpod_cli/custom_launcher/launcher/config_validator/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/src/hyperpod_cli/custom_launcher/launcher/config_validator/type_validator.py b/src/hyperpod_cli/custom_launcher/launcher/config_validator/type_validator.py
deleted file mode 100644
index 2daa6dd..0000000
--- a/src/hyperpod_cli/custom_launcher/launcher/config_validator/type_validator.py
+++ /dev/null
@@ -1,137 +0,0 @@
-import os
-from typing import Any
-
-from .value_validator import get_argument
-from omegaconf import DictConfig, OmegaConf
-
-
-class TypeValidator:
-    # Define the list of key-type pairs
-    types = [
-        ('hydra.output_subdir', 'path'),
-        ('hydra.run.dir', 'path'),
-        ('git.repo_url', 'string'),
-        ('git.branch', 'string'),
-        ('git.commit', 'string'),
-        ('training_cfg.entry_script', 'path'),
-        ('training_cfg.script_args', 'list_dict'),
-        ('training_cfg.run.name', 'string'),
-        ('training_cfg.run.nodes', 'positive_integer'),
-        ('training_cfg.run.ntasks_per_node', 'positive_integer'),
-        ('cluster.cluster_type', 'string'),
-        ('cluster.instance_type', 'string'),
-        ('cluster.cluster_config', 'dict'),
-        ('cluster.cluster_config.namespace', 'string'),
-        ('cluster.cluster_config.custom_labels', 'dict'),
-        ('cluster.cluster_config.annotations', 'dict'),
-        ('cluster.cluster_config.priority_class_name', 'string'),
-        ('cluster.cluster_config.label_selector', 'dict'),
-        ('cluster.cluster_config.persistentVolumeClaims', 'list_dict'),
-        ('cluster.cluster_config.volumes', 'list_dict'),
-        ('cluster.cluster_config.pullPolicy', 'string'),
-        ('cluster.cluster_config.restartPolicy', 'string'),
-        ('base_results_dir', 'path'),
-        ('container_mounts', 'list_path'),
-        ('container', 'string'),
-        ('env_vars', 'dict')
-    ]
-
-    def __init__(self, config: DictConfig):
-        self.config = config
-
-    def validate(self):
-        for key, type in self.types:
-            argument = get_argument(self.config, key)
-            _check_types(argument, type, key)
-
-
-def _is_valid_path(path) -> bool:
-    """
-    Check if the input string is a valid file path.
-
-    Parameters:
-    path (str): The path to validate.
-
-    Returns:
-    bool: True if the path is valid, False otherwise.
-    """
-    if not isinstance(path, str):
-        return False
-
-    try:
-        normalized_path = os.path.normpath(path)
-        return True
-    except Exception as e:
-        return False
-
-
-def _is_positive_integer(argument) -> bool:
-    try:
-        val = int(argument)
-        if val < 1:
-            return False
-    except ValueError:
-        return False
-    return True
-
-
-def _get_base_omega_conf_container(argument) -> (Any, bool):
-    try:
-        argument = OmegaConf.to_container(argument, resolve=True)
-    except Exception as e:
-        return None, False
-    return argument,True
-
-
-def _is_list_of_dicts(argument) -> bool:
-    argument, status = _get_base_omega_conf_container(argument)
-    if status is False:
-        return False
-    return isinstance(argument, list) and all(item is None or isinstance(item, dict) for item in argument)
-
-
-def _is_list_of_strings(argument) -> bool:
-    argument, status = _get_base_omega_conf_container(argument)
-    if status is False:
-        return False
-    return isinstance(argument, list) and all(item is None or isinstance(item, str) for item in argument)
-
-
-def _is_list_of_paths(argument) -> bool:
-    argument, status = _get_base_omega_conf_container(argument)
-    if status is False:
-        return False
-    return isinstance(argument, list) and all(item is None or _is_valid_path(item) for item in argument)
-
-
-def _is_dict(argument) -> bool:
-    argument, status = _get_base_omega_conf_container(argument)
-    if status is False:
-        return False
-    return isinstance(argument, dict)
-
-
-def _check_types(argument, type, argument_name) -> None:
-    if argument is None:
-        return
-
-    if type == "string" and not isinstance(argument, str):
-        raise TypeError("{} with val {} is not a string".format(argument_name, argument))
-
-    if type == "path" and not _is_valid_path(argument):
-        raise TypeError("{} with val {} is not a valid path".format(argument_name, argument))
-
-    if type == "list_string" and not _is_list_of_strings(argument):
-        raise TypeError("{} with val {} is not a list of string".format(argument_name, argument))
-
-    if type == "list_dict" and not _is_list_of_dicts(argument):
-        raise TypeError("{} with val {} is not a list of dictionary".format(argument_name, argument))
-
-    if type == "list_path" and not _is_list_of_paths(argument):
-        raise TypeError("{} with val {} is not a list of paths".format(argument_name, argument))
-
-    if type == "positive_integer" and not _is_positive_integer(argument):
-        raise TypeError("{} with val {} is not a positive integer".format(argument_name, argument))
-
-    if type == "dict" and not _is_dict(argument):
-        raise TypeError("{} with val {} is not a dictionary".format(argument_name, argument))
diff --git a/src/hyperpod_cli/custom_launcher/launcher/config_validator/value_validator.py b/src/hyperpod_cli/custom_launcher/launcher/config_validator/value_validator.py
deleted file mode 100644
index 9252d1a..0000000
--- a/src/hyperpod_cli/custom_launcher/launcher/config_validator/value_validator.py
+++ /dev/null
@@ -1,216 +0,0 @@
-import re
-
-from omegaconf import DictConfig
-
-
-class ValueValidator:
-    def __init__(self, config: DictConfig):
-        self.config = config
-
-    def validate(self) -> None:
-        # For all below validations, we will check if the argument is present => it should pass the validation
-
-        # Mandatory arguments for all workflows
-        _validate_all_mandatory_argument(self.config)
-
-        # Cluster type argument check for all workflows
-        _validate_cluster_type_argument(self.config)
-
-        # PV argument check for all workflows
-        _validate_pv_arguments(self.config)
-
-        # Volume argument check for all workflows
-        _validate_volume_arguments(self.config)
-
-        # Pull policy argument check for all workflows
-        _validate_pull_policy_argument(self.config)
-
-        # Restart policy argument check for all workflows
-        _validate_restart_policy_argument(self.config)
-
-        # Namespace argument regex check for k8 workflows
-        _validate_namespace_argument(self.config)
-
-        # Check for mandatory arguments for k8 custom script workflow
-        _validate_k8_custom_script_workflow_mandatory_argument(self.config)
-
-
-def _validate_mandatory_argument(argument, argument_name: str) -> None:
-    """
-    Check if the mandatory input argument is not None.
-
-    Parameters:
-    argument : The argument to validate.
-    argument_name : The name of argument to validate.
-    """
-    if argument is None:
-        raise ValueError("Missing mandatory argument " + argument_name + " is not provided")
-
-
-def get_argument(config: DictConfig, argument_name: str):
-    """
-    Return the value of given argument_name from config
-
-    Parameters:
-    config : Configuration dictionary
-    argument_name : The name of argument to fetch.
-
-    Returns:
-    bool: Argument if present else return None
-    argument_name can have a nested key like key1.key2 in which case we will return config[key1][key2].
-    If  config[key1] is None for above case we will return None
-    """
-    argument_name_splits = argument_name.split('.')
-    if len(argument_name_splits) > 1:
-        subconfig = config.get(argument_name_splits[0])
-        if subconfig is None:
-            return None
-        argument_name_splits.pop(0)
-        remaining_argument_name = '.'.join(argument_name_splits)
-        return get_argument(subconfig, remaining_argument_name)
-    return config.get(argument_name)
-
-
-def _validate_pv_arguments(config: DictConfig) -> None:
-    """
-    Check all the information needed for persistentVolumeClaim is provided if it is not None
-
-    Parameters:
-    config (DictConfig): Configuration dictionary
-    """
-    cluster_config_name = "cluster.cluster_config"
-    cluster_config = get_argument(config, cluster_config_name)
-    pv_argument_name = "persistentVolumeClaims"
-    exception_message: str = "claimName and mountPath should be provided for persistentVolumeClaim"
-    if cluster_config is not None and pv_argument_name in cluster_config:
-        pv_arguments = cluster_config.get(pv_argument_name)
-        claim_name = "claimName"
-        mount_path = "mountPath"
-        for pv_argument in pv_arguments:
-            if pv_argument is None or claim_name not in pv_argument or mount_path not in pv_argument:
-                raise ValueError(exception_message)
-            claim_name_argument = pv_argument.get(claim_name)
-            mount_path_argument = pv_argument.get(mount_path)
-            if claim_name_argument is None or mount_path_argument is None:
-                raise ValueError(exception_message)
-
-def _validate_volume_arguments(config: DictConfig) -> None:
-    """
-    Check all the information needed for volume is provided if it is not None
-
-    Parameters:
-    config (DictConfig): Configuration dictionary
-    """
-    cluster_config_name = "cluster.cluster_config"
-    cluster_config = get_argument(config, cluster_config_name)
-    volumes_argument_name = "volumes"
-    exception_message: str = "hostPath, mountPath, volumeName should be provided for volumes"
-    if cluster_config is not None and volumes_argument_name in cluster_config:
-        volume_arguments = cluster_config.get(volumes_argument_name)
-        if volume_arguments is None:
-            return
-        host_path = "hostPath"
-        mount_path = "mountPath"
-        volume_name = "volumeName"
-        for volume_argument in volume_arguments:
-            if volume_argument is None or host_path not in volume_argument or mount_path not in volume_argument or volume_name not in volume_argument:
-                raise ValueError(exception_message)
-            host_path_argument = volume_argument.get(host_path)
-            mount_path_argument = volume_argument.get(mount_path)
-            volume_name_argument = volume_argument.get(volume_name)
-            if host_path_argument is None or mount_path_argument is None or volume_name_argument is None:
-                raise ValueError(exception_message)
-
-def _validate_pull_policy_argument(config: DictConfig) -> None:
-    """
-    Check only valid pullPolicy is provided if it is not None
-
-    Parameters:
-    config (DictConfig): Configuration dictionary
-    """
-    pull_policy_argument_name = "cluster.cluster_config.pullPolicy"
-    pull_policy_argument = get_argument(config, pull_policy_argument_name)
-    if pull_policy_argument is not None:
-        supported_pull_policies = ["Always", "IfNotPresent", "Never"]
-        if pull_policy_argument not in supported_pull_policies:
-            raise ValueError("Provided pullPolicy is not supported")
-
-
-def _validate_restart_policy_argument(config: DictConfig) -> None:
-    """
-    Check only valid restartPolicy is provided if it is not None
-
-    Parameters:
-    config (DictConfig): Configuration dictionary
-    """
-    restart_policy_argument_name = "cluster.cluster_config.restartPolicy"
-    restart_policy_argument = get_argument(config, restart_policy_argument_name)
-    if restart_policy_argument is not None:
-        supported_restart_policies = ["Always", "OnFailure", "Never", "ExitCode"]
-        if restart_policy_argument not in supported_restart_policies:
-            raise ValueError("Provided restartPolicy is not supported")
-
-
-def _validate_cluster_type_argument(config: DictConfig) -> None:
-    """
-    Check only valid cluster_type is provided
-
-    Parameters:
-    config (DictConfig): Configuration dictionary
-    """
-    cluster_type_argument_name = "cluster.cluster_type"
-    cluster_type = get_argument(config, cluster_type_argument_name)
-    supported_cluster_types = ["k8s"]
-    if cluster_type is not None and cluster_type not in supported_cluster_types:
-        raise ValueError("Provided cluster_type is not supported")
-
-
-def _validate_namespace_argument(config: DictConfig) -> None:
-    """
-    Check only valid kubectl namespace is provided
-    Naming Convention of Kubernetes Namespaces is
-    You can create a name with a maximum length of 253 characters using only alphanumeric characters and hyphens.
-    Names cannot start with a hyphen and the alpha characters can only be lowercase.
-
-    Parameters:
-    config (DictConfig): Configuration dictionary
-    """
-
-    """
-    Here's a breakdown of the regex pattern:
-
-    ^ - Asserts the position at the start of the string.
-    (?!-) - A negative lookahead assertion to ensure the string does not start with a hyphen.
-    [a-z0-9-] - Matches any lowercase letter, digit, or hyphen.
-    {1,253} - Specifies that the string must be between 1 and 253 characters long.
-    $ - Asserts the position at the end of the string.
-    This pattern will ensure the string is constructed using only lowercase alphanumeric characters and hyphens, 
-    is no longer than 253 characters, and does not start with a hyphen.
-    """
-    namespace_regex = r'^(?!-)[a-z0-9-]{1,253}$'
-    namespace_argument_name = "cluster.cluster_config.namespace"
-    namespace_argument = get_argument(config, namespace_argument_name)
-    if namespace_argument is not None and not re.match(namespace_regex, namespace_argument):
-        raise ValueError("Provided namespace is not valid, Kindly provide Kubernetes Namespace "
-                         "with a maximum length of 253 characters using only alphanumeric characters and hyphens. "
-                         "Names cannot start with a hyphen and the alpha characters can only be lowercase.")
-
-
-def _validate_all_mandatory_argument(config: DictConfig) -> None:
-    mandatory_arguments_for_all_workflows = ["base_results_dir"]
-    for argument_name in mandatory_arguments_for_all_workflows:
-        argument = get_argument(config, argument_name)
-        _validate_mandatory_argument(argument, argument_name)
-
-
-def _validate_k8_custom_script_workflow_mandatory_argument(config: DictConfig) -> None:
-    cluster_type = get_argument(config, "cluster.cluster_type")
-    training_cfg = get_argument(config, "training_cfg")
-    if cluster_type == "k8s" and training_cfg is not None:
-        k8_custom_script_mandatory_arguments = ["container", "env_vars",
-                                                "training_cfg", "training_cfg.entry_script",
-                                                "training_cfg.run", "training_cfg.run.name", "training_cfg.run.nodes",
-                                                "cluster", "cluster.cluster_type", "cluster.instance_type", "cluster.cluster_config"]
-        for argument_name in k8_custom_script_mandatory_arguments:
-            argument = get_argument(config, argument_name)
-            _validate_mandatory_argument(argument, argument_name)
diff --git a/src/hyperpod_cli/custom_launcher/launcher/efa.py b/src/hyperpod_cli/custom_launcher/launcher/efa.py
deleted file mode 100644
index fd0a354..0000000
--- a/src/hyperpod_cli/custom_launcher/launcher/efa.py
+++ /dev/null
@@ -1,142 +0,0 @@
-efa_supported_instance = set(
-    [
-        "c5n.18xlarge",
-        "c5n.9xlarge",
-        "c5n.metal",
-        "c6a.48xlarge",
-        "c6a.metal",
-        "c6gn.16xlarge",
-        "c6i.32xlarge",
-        "c6i.metal",
-        "c6id.32xlarge",
-        "c6id.metal",
-        "c6in.32xlarge",
-        "c6in.metal",
-        "c7a.48xlarge",
-        "c7a.metal-48xl",
-        "c7g.16xlarge",
-        "c7g.metal",
-        "c7gd.16xlarge",
-        "c7gd.metal",
-        "c7gn.16xlarge",
-        "c7gn.metal",
-        "c7i.48xlarge",
-        "c7i.metal-48xl",
-        "dl1.24xlarge",
-        "dl2q.24xlarge",
-        "g4dn.12xlarge",
-        "g4dn.16xlarge",
-        "g4dn.8xlarge",
-        "g4dn.metal",
-        "g5.12xlarge",
-        "g5.16xlarge",
-        "g5.24xlarge",
-        "g5.48xlarge",
-        "g5.8xlarge",
-        "g6.12xlarge",
-        "g6.16xlarge",
-        "g6.24xlarge",
-        "g6.48xlarge",
-        "g6.8xlarge",
-        "gr6.8xlarge",
-        "i3en.12xlarge",
-        "i3en.24xlarge",
-        "i3en.metal",
-        "i4g.16xlarge",
-        "i4i.32xlarge",
-        "i4i.metal",
-        "im4gn.16xlarge",
-        "inf1.24xlarge",
-        "m5dn.24xlarge",
-        "m5dn.metal",
-        "m5n.24xlarge",
-        "m5n.metal",
-        "m5zn.12xlarge",
-        "m5zn.metal",
-        "m6a.48xlarge",
-        "m6a.metal",
-        "m6i.32xlarge",
-        "m6i.metal",
-        "m6id.32xlarge",
-        "m6id.metal",
-        "m6idn.32xlarge",
-        "m6idn.metal",
-        "m6in.32xlarge",
-        "m6in.metal",
-        "m7a.48xlarge",
-        "m7a.metal-48xl",
-        "m7g.16xlarge",
-        "m7g.metal",
-        "m7gd.16xlarge",
-        "m7gd.metal",
-        "m7i.48xlarge",
-        "m7i.metal-48xl",
-        "p3dn.24xlarge",
-        "p4d.24xlarge",
-        "p4de.24xlarge",
-        "p5.48xlarge",
-        "r5dn.24xlarge",
-        "r5dn.metal",
-        "r5n.24xlarge",
-        "r5n.metal",
-        "r6a.48xlarge",
-        "r6a.metal",
-        "r6i.32xlarge",
-        "r6i.metal",
-        "r6id.32xlarge",
-        "r6id.metal",
-        "r6idn.32xlarge",
-        "r6idn.metal",
-        "r6in.32xlarge",
-        "r6in.metal",
-        "r7a.48xlarge",
-        "r7a.metal-48xl",
-        "r7g.16xlarge",
-        "r7g.metal",
-        "r7gd.16xlarge",
-        "r7gd.metal",
-        "r7i.48xlarge",
-        "r7i.metal-48xl",
-        "r7iz.32xlarge",
-        "r7iz.metal-32xl",
-        "r8g.24xlarge",
-        "r8g.48xlarge",
-        "r8g.metal-24xl",
-        "r8g.metal-48xl",
-        "trn1.32xlarge",
-        "trn1n.32xlarge",
-        "u7i-12tb.224xlarge",
-        "u7in-16tb.224xlarge",
-        "u7in-24tb.224xlarge",
-        "u7in-32tb.224xlarge",
-        "vt1.24xlarge",
-        "x2idn.32xlarge",
-        "x2idn.metal",
-        "x2iedn.32xlarge",
-        "x2iedn.metal",
-        "x2iezn.12xlarge",
-        "x2iezn.metal",
-    ]
-)
-
-instanceWithMultipleEFAs = {
-    "p4d.24xlarge": 4,
-    "p4de.24xlarge": 4,
-    "p5.4xlarge": 4,
-    "p5.24xlarge": 16,
-    "p5.48xlarge": 32,
-    "trn1.32xlarge": 8,
-    "trn1n.32xlarge": 16,
-}
-
-instanceWithRDMASupport = set(
-    [
-        "p4d.24xlarge",
-        "p4de.24xlarge",
-        "p5.4xlarge",
-        "p5.24xlarge",
-        "p5.48xlarge",
-        "trn1.32xlarge",
-        "trn1n.32xlarge",
-    ]
-)
diff --git a/src/hyperpod_cli/custom_launcher/launcher/nemo/__init__.py b/src/hyperpod_cli/custom_launcher/launcher/nemo/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/src/hyperpod_cli/custom_launcher/launcher/nemo/k8s_templates/training/Chart.yaml b/src/hyperpod_cli/custom_launcher/launcher/nemo/k8s_templates/training/Chart.yaml
deleted file mode 100644
index 5665246..0000000
--- a/src/hyperpod_cli/custom_launcher/launcher/nemo/k8s_templates/training/Chart.yaml
+++ /dev/null
@@ -1,5 +0,0 @@
-apiVersion: v2
-appVersion: "1.0"
-description: Sagemaker Model Training
-name: sagemaker-training
-version: 1.0.0
diff --git a/src/hyperpod_cli/custom_launcher/launcher/nemo/k8s_templates/training/training-config.yaml b/src/hyperpod_cli/custom_launcher/launcher/nemo/k8s_templates/training/training-config.yaml
deleted file mode 100644
index ce30951..0000000
--- a/src/hyperpod_cli/custom_launcher/launcher/nemo/k8s_templates/training/training-config.yaml
+++ /dev/null
@@ -1,7 +0,0 @@
-apiVersion: v1
-kind: ConfigMap
-metadata:
-  name: training-config
-data:
-  config.yaml: |-
-  {{ (.Files.Glob "config/*hydra.yaml").AsConfig | indent 4 }}
diff --git a/src/hyperpod_cli/custom_launcher/launcher/nemo/k8s_templates/training/training.yaml b/src/hyperpod_cli/custom_launcher/launcher/nemo/k8s_templates/training/training.yaml
deleted file mode 100644
index 9377271..0000000
--- a/src/hyperpod_cli/custom_launcher/launcher/nemo/k8s_templates/training/training.yaml
+++ /dev/null
@@ -1,227 +0,0 @@
-{{ $config := .Values.trainingConfig }}
-apiVersion: v1
-kind: ConfigMap
-metadata:
-  name: train-script-{{ $config.jobName }}
-data:
-  train-script.sh: |
-    #!/bin/bash
-    set -ex
-
-    {{- if $config.git.repo_url }}
-    mkdir -p $HOME/tmp
-    GIT_CLONE_DIR=$HOME/tmp/$HOSTNAME
-    [[ -d $GIT_CLONE_DIR ]] && rm -rf $GIT_CLONE_DIR
-    git clone {{ $config.git.repo_url }} $GIT_CLONE_DIR
-    GIT_CLONE_DIR=${GIT_CLONE_DIR}/
-    cd $GIT_CLONE_DIR
-
-    {{- if $config.git.branch }}
-    git checkout {{ $config.git.branch }}
-    {{- end }}
-
-    {{- if $config.git.commit }}
-    git fetch origin {{ $config.git.commit }}
-    git reset --hard {{ $config.git.commit }}
-    {{- end }}
-    {{- else }}
-    GIT_CLONE_DIR=""
-    {{- end }}
-
-    {{- range $config.pre_script }}
-    {{ . }}
-    {{- end }}
-
-    {{- if gt (int $config.nodes) 1 }}
-    {{- if eq $config.device "trainium" }}
-    hostname=$(hostname)
-    prefix="{{ $config.jobName }}-worker-"
-    echo "prefix is $prefix"
-    node_id=${hostname#"$prefix"}
-    export DISTRIBUTED_ARGS="--nproc_per_node {{ $config.ntasksPerNode }} --nnodes {{ $config.nodes }} --node_rank $node_id --master_addr={{ $config.jobName }}-worker-0 --master_port 44000"
-    {{- else }}
-    export DISTRIBUTED_ARGS="--nproc_per_node {{ $config.ntasksPerNode }} --nnodes {{ $config.nodes }} --rdzv_backend=c10d --rdzv_endpoint={{ $config.jobName }}-worker-0"
-    {{- end }}
-    {{- else }}
-    export DISTRIBUTED_ARGS="--nproc_per_node {{ $config.ntasksPerNode }}"
-    {{- end }}
-
-    echo "DISTRIBUTED_ARGS=$DISTRIBUTED_ARGS"
-    torchrun $DISTRIBUTED_ARGS ${GIT_CLONE_DIR}{{ $config.scriptPath }} \
-    {{- if $config.scriptArgs }}
-    {{ $config.scriptArgs }}
-    {{- end }}
-
-    {{- range $config.post_script }}
-    {{ . }}
-    {{- end }}
-
-    {{- if $config.git.repo_url }}
-    cd $HOME
-    rm -rf $GIT_CLONE_DIR
-    {{- end }}
----
-apiVersion: kubeflow.org/v1
-kind: PyTorchJob
-metadata:
-  name: {{ $config.jobName }}
-  namespace: {{ $config.namespace }}
-  {{- if $config.annotations }}
-  annotations:
-    {{- range $key, $value := $config.annotations }}
-    {{ $key | quote }}: {{ $value | quote }}
-    {{- end }}
-  {{- end }}
-  labels:
-    app: {{ $config.jobName }}
-    {{- if $config.customLabels }}
-    {{- range $key, $value := $config.customLabels }}
-    {{ $key | quote }}: {{ $value | quote }}
-    {{- end}}
-    {{- end }}
-spec:
-  pytorchReplicaSpecs:
-    Worker:
-      replicas: {{ $config.nodes }}
-      template:
-        {{- if $config.customLabels }}
-        metadata:
-          labels:
-            {{- range $key, $value := $config.customLabels }}
-            {{ $key | quote }}: {{ $value | quote }}
-            {{- end }}
-        {{- end }}
-        spec:
-          {{- if $config.priorityClassName }}
-          priorityClassName: {{ $config.priorityClassName }}
-          {{- end }}
-          {{- if $config.serviceAccountName }}
-          serviceAccountName: {{ $config.serviceAccountName }}
-          {{- end }}
-          containers:
-          - name: pytorch
-            image: {{ .Values.image.trainingImage }}
-            env:
-              {{- range $key, $value := $config.envVars }}
-              - name: {{ $key }}
-                value: {{ $value | quote }}
-              {{- end}}
-            command:
-            - /etc/config/train-script.sh
-            imagePullPolicy: {{ .Values.image.pullPolicy }}
-            securityContext:
-              capabilities:
-                add: [ "IPC_LOCK" ]
-            {{- if or (eq $config.device "gpu") (eq $config.device "trainium") (gt (int $config.numEFADevices) 0 ) }}
-            resources:
-              requests:
-                {{- if eq $config.device "gpu" }}
-                nvidia.com/gpu: {{ $config.ntasksPerNode }}
-                {{- end }}
-                {{- if eq $config.device "trainium" }}
-                aws.amazon.com/neurondevice: {{ $config.numNeuronDevices }}
-                {{- end }}
-                {{- if gt (int $config.numEFADevices) 0 }}
-                vpc.amazonaws.com/efa: {{ $config.numEFADevices }}
-                {{- end }}
-              limits:
-                {{- if eq $config.device "gpu" }}
-                nvidia.com/gpu: {{ $config.ntasksPerNode }}
-                {{- end }}
-                {{- if eq $config.device "trainium" }}
-                aws.amazon.com/neurondevice: {{ $config.numNeuronDevices }}
-                {{- end }}
-                {{- if gt (int $config.numEFADevices) 0 }}
-                vpc.amazonaws.com/efa: {{ $config.numEFADevices }}
-                {{- end }}
-            {{- end }}
-            volumeMounts:
-            {{- if $config.persistentVolumeClaims }}
-            {{- range $config.persistentVolumeClaims }}
-            - mountPath: {{ .mountPath }}
-              name: {{ .claimName }}-volume
-            {{- end }}
-            {{- end }}
-            {{- if $config.volumes }}
-            {{- range $config.volumes }}
-            - mountPath: {{ .mountPath }}
-              name: {{ .volumeName }}
-            {{- end }}
-            {{- end }}            
-            {{- if not $config.customScript }}
-            - mountPath: /config
-              name: training-config
-            {{- end }}
-            - mountPath: /etc/config
-              name: train-script
-            - mountPath: /dev/shm
-              name: shm
-          restartPolicy: {{ $config.restartPolicy }}
-
-          {{- if (or $config.labelSelector.required $config.labelSelector.preferred) }}
-          affinity:
-            nodeAffinity:
-            {{- if $config.labelSelector.required }}
-              {{- range $key, $values := $config.labelSelector.required }}
-              requiredDuringSchedulingIgnoredDuringExecution:
-                nodeSelectorTerms:
-                  - matchExpressions:
-                    - key: {{ $key | quote }}
-                      operator: In
-                      values:
-                        {{- range $values }}
-                        - {{ . | quote }}
-                        {{- end}}
-              {{- end }}
-            {{- end }}
-
-            {{- if $config.labelSelector.preferred }}
-              {{- $index := 0 }}
-              {{- range $key, $values := $config.labelSelector.preferred }}
-              preferredDuringSchedulingIgnoredDuringExecution:
-                - weight: {{ index $config.labelSelector.weights $index }}
-                  preference:
-                    matchExpressions:
-                      - key: {{ $key | quote }}
-                        operator: In
-                        values:
-                          {{- range $values }}
-                          - {{ . | quote }}
-                          {{- end }}
-              {{- $index = add $index 1 }}
-              {{- end }}
-            {{- end }}
-          {{- end }}
-
-          volumes:
-          {{- if $config.persistentVolumeClaims }}
-          {{- range $config.persistentVolumeClaims }}
-          - name: {{ .claimName }}-volume
-            persistentVolumeClaim:
-              claimName: {{ .claimName }}
-          {{- end }}
-          {{- end }}
-          {{- if $config.volumes }}
-          {{- range $config.volumes }}
-          - name: {{ .volumeName }}
-            hostPath:
-              path: {{ .hostPath }}
-          {{- end }}
-          {{- end }}          
-          {{- if not $config.customScript }}
-          - configMap:
-              name: training-config
-            name: training-config
-          {{- end }}
-          - name: shm
-            hostPath:
-              path: /dev/shm
-              type: Directory
-          - name: train-script
-            configMap:
-              defaultMode: 420
-              items:
-              - key: train-script.sh
-                mode: 365
-                path: train-script.sh
-              name: train-script-{{ $config.jobName }}
diff --git a/src/hyperpod_cli/custom_launcher/launcher/nemo/k8s_templates/training/values.yaml b/src/hyperpod_cli/custom_launcher/launcher/nemo/k8s_templates/training/values.yaml
deleted file mode 100644
index fc432d2..0000000
--- a/src/hyperpod_cli/custom_launcher/launcher/nemo/k8s_templates/training/values.yaml
+++ /dev/null
@@ -1,77 +0,0 @@
-image:
-  # training image
-  trainingImage: cfg.container
-
-  # image pulling policy
-  pullPolicy: IfNotPresent
-
-
-trainingConfig:
-  # current job name
-  jobName: "nil"
-
-  # namespace to launch job
-  namespace: "default"
-
-  # script path
-  scriptPath: null
-
-  # script args
-  scriptArgs: null
-
-  # specify whether to use custom scripts
-  customScript: null
-
-  # list of custom annotations apply to jobs
-  annotations: null
-
-  # list of custom labels apply to jobs and pods
-  customLabels: null
-
-  # Kueue scheduler priority class name
-  priority_class_name: null
-
-  # device type, can be "gpu", "trainium" and "nil", "nil" means cpu
-  device: "nil"
-
-  # number of EFA devices if the instance type support EFA
-  numEFADevices: 0
-
-  # number of process per node
-  ntasksPerNode: 0
-
-  # number of nodes to run
-  nodes: training.trainer.num_nodes
-
-  # restart policy
-  restartPolicy: Never
-
-  # from NeMo, not used currently
-  wandbKey: "nil"
-
-  # name of service account associated with the namespace
-  serviceAccountName: null
-
-  # persistent volume, usually used to mount FSx
-  persistentVolumeClaims: null
-
-  # temp volume mount
-  volumes: null
-
-  # A github repo if user might want to use script inside
-  git:
-    repo_url: null
-    branch: null
-    commit: null
-    token: null
-
-  # Commands to run before training
-  pre_script: []
-  # Commands to run after training
-  post_script: []
-
-  # select preferred and required labels for nodes
-  labelSelector:
-    required: null # select nodes with required labels
-    preferred: null # select nodes with priority which has preferred labels
-    weights: null # list of weights for the preferred labels
diff --git a/src/hyperpod_cli/custom_launcher/launcher/nemo/launchers.py b/src/hyperpod_cli/custom_launcher/launcher/nemo/launchers.py
deleted file mode 100755
index 3742931..0000000
--- a/src/hyperpod_cli/custom_launcher/launcher/nemo/launchers.py
+++ /dev/null
@@ -1,58 +0,0 @@
-from pathlib import Path
-from typing import List
-
-import nemo_launcher.utils.job_utils as job_utils
-from nemo_launcher.core.launchers import AutoLauncher, K8SLauncher
-
-
-
-class SMAutoLauncher(AutoLauncher):
-    """
-    AutoLauncher object for Sagemaker
-    """
-
-    @staticmethod
-    def get_launchers():
-        """Returns supported launchers as a dictionary from launcher name to launcher class"""
-        return {
-            "k8s": SMK8SLauncher,
-        }
-
-
-class SMJobPaths(job_utils.JobPaths):
-    """
-    Our launcher contains an extra entry script called train_script.sh
-    This class is used to specify its path
-    """
-
-    @property
-    def train_script_file(self) -> Path:
-        return self._folder / f"train_script.sh"
-
-
-class SMK8SLauncher(K8SLauncher):
-    """
-    Launcher for SM training jobs using K8s.
-    """
-
-    def _make_submission_file_text(self, command_groups: List[List[str]]) -> str:
-        """
-        Generate the script to launch the Helm chart.
-        A very simple bash script is generated which runs `helm install` for the
-        Helm chart that was generated.
-
-        :param List[List[str]] command_groups: Command groups to launch with
-        :return: submission script file's text
-        :rtype: str
-        """
-        paths = job_utils.JobPaths(folder=self.folder, job_name=self.job_name)
-        helm_charts = paths.folder / "k8s_template"
-        job_name = self.job_name.replace("_", "-")
-
-        extra_helm_args = ""
-        if self.parameters.get("namespace", None):
-            extra_helm_args += f" --namespace {self.parameters['namespace']}"
-
-        # Apply a timeout of 15min in case images take a long time to bring up
-        # or pre-install hooks take a while
-        return f"#!/bin/bash\nhelm install --timeout=15m --wait {extra_helm_args} {job_name} {helm_charts}\n"
diff --git a/src/hyperpod_cli/custom_launcher/launcher/nemo/nemo_framework_launcher b/src/hyperpod_cli/custom_launcher/launcher/nemo/nemo_framework_launcher
deleted file mode 160000
index 3d41c31..0000000
--- a/src/hyperpod_cli/custom_launcher/launcher/nemo/nemo_framework_launcher
+++ /dev/null
@@ -1 +0,0 @@
-Subproject commit 3d41c31c91d5a47a84ad15cbf783c56700c30521
diff --git a/src/hyperpod_cli/custom_launcher/launcher/nemo/stages.py b/src/hyperpod_cli/custom_launcher/launcher/nemo/stages.py
deleted file mode 100755
index 8c1e9e3..0000000
--- a/src/hyperpod_cli/custom_launcher/launcher/nemo/stages.py
+++ /dev/null
@@ -1,573 +0,0 @@
-import os
-import shutil
-from pathlib import Path
-from typing import Dict, List
-
-import omegaconf
-from urllib.parse import urlparse
-from nemo_launcher.core.stages import Training
-from nemo_launcher.utils.job_utils import JobPaths
-from omegaconf import OmegaConf
-
-from ..accelerator_devices import get_num_accelerator_devices
-from ..efa import (
-    efa_supported_instance,
-    instanceWithMultipleEFAs,
-    instanceWithRDMASupport,
-)
-from .launchers import SMAutoLauncher
-
-# Predefined distributed args for torchrun
-PROCESSES_PER_NODE = "PROCESSES_PER_NODE"
-NNODES = "NNODES"
-NODEID = "NODEID"
-MASTER_ADDR = "MASTER_ADDR"
-MASTER_PORT = "MASTER_PORT"
-DISTRIBUTED_ARGS = "DISTRIBUTED_ARGS"
-
-
-def set_multinode_envs(env_vars, instance_type):
-    # https://github.com/aws/aws-ofi-nccl/blob/master/doc/efa-env-var.md
-    if get_num_efa_devices(instance_type) > 0:
-        env_vars["FI_PROVIDER"] = "efa"
-        env_vars["NCCL_PROTO"] = "simple"
-        if allow_rdma(instance_type):
-            env_vars["FI_EFA_USE_DEVICE_RDMA"] = "1"
-    env_vars["NCCL_SOCKET_IFNAME"] = "^lo,docker0"
-    env_vars["NCCL_IGNORE_DISABLED_P2P"] = "1"
-    env_vars["TORCH_NCCL_ASYNC_ERROR_HANDLING"] = "1"
-    env_vars["TORCH_DIST_INIT_BARRIER"] = "1"
-    env_vars["CUDA_DEVICE_MAX_CONNECTIONS"] = "1"
-    env_vars["FI_EFA_FORK_SAFE"] = "1"
-    return env_vars
-
-
-def allow_rdma(instance_type):
-    return instance_type in instanceWithRDMASupport
-
-
-def get_instance_type(cfg):
-    instance_type = None
-    if cfg.get("instance_type"):
-        instance_type = cfg.instance_type
-    else:
-        # custom path
-        instance_type = cfg.cluster.instance_type
-    assert instance_type is not None, "instance type is required from config"
-    if instance_type.startswith("ml."):
-        instance_type = instance_type[3:]
-    return instance_type.lower()
-
-
-def get_num_efa_devices(instance_type):
-    # If not a EFA instance, return 0
-    if instance_type not in efa_supported_instance:
-        return 0
-    # If multi-EFA, return from mapping
-    if instance_type in instanceWithMultipleEFAs:
-        return instanceWithMultipleEFAs[instance_type]
-    # Only a single EFA device
-    return 1
-
-
-def get_ntasks_per_node(stage_cfg):
-    """
-    Get the number of processes per node used for training
-    When running with custom script it will be stage_cfg.run.ntasks_per_node
-    """
-    run_cfg = stage_cfg.get("run")
-    ntasks = run_cfg.get("ntasks_per_node")
-    if ntasks is None:
-        ntasks = stage_cfg.get("trainer").get("devices")
-    return ntasks
-
-
-def get_num_nodes(stage_cfg):
-    """
-    Get the number of nodes used for training
-    When running with custom script it will be stage_cfg.run.nodes
-    """
-    run_cfg = stage_cfg.get("run")
-    nodes = run_cfg.get("nodes")
-    if nodes is None:
-        nodes = stage_cfg.get("trainer").get("num_nodes")
-    return nodes
-
-
-class SMTraining(Training):
-    """
-    Base stage class for doing training on Sagemaker
-    """
-
-    def __init__(self, cfg):
-        super().__init__(cfg)
-        # Use GPU device for default flow for NeMo runs
-        self.device = "gpu"
-        self.instance_type = get_instance_type(cfg)
-        self.num_efa_devices = get_num_efa_devices(self.instance_type)
-
-    @property
-    def _default_repo(self):
-        # Default repo to mount script from
-        return None
-
-    def _make_torchrun_string(self):
-        """
-        Create torchrun string based on single/multi-node job
-        """
-        ntasks_per_node = get_ntasks_per_node(self.stage_cfg)
-        if int(get_num_nodes(self.stage_cfg)) > 1:
-            return f"torchrun ${DISTRIBUTED_ARGS} "
-        else:
-            return f"torchrun --nproc_per_node {ntasks_per_node} "
-
-    def _make_custom_call_string(self, stage_cfg_path=None):
-        """
-        Create the training command with torchrun, script and args
-        """
-        script_path = str(self._entry_script_path)
-        torchrun_cmd = self._make_torchrun_string()
-        script_args_str = self.get_script_args_str(stage_cfg_path)
-        command = [torchrun_cmd, script_path, script_args_str]
-        command_string = " \\\n  ".join(command)
-        return command_string
-
-    def _get_hostfile_location(self):
-        """
-        Get the file location to store the hostnames
-        """
-        job_path = self.get_job_path()
-        hostfile_location = Path(job_path.folder / "hostname")
-        return hostfile_location
-
-    def _get_nodeid_location(self):
-        """
-        Get the file location to store the nodeid, which will be used for node_rank in torchrun arg
-        """
-        job_path = self.get_job_path()
-        nodeid_location = Path(job_path.folder / "node_id")
-        return nodeid_location
-
-    def _make_train_script_text(self, stage_cfg_path=None, port=41000) -> str:
-        """
-        The custom train entry script, it will be responsible for following
-        - Handle resolving hostname and create torch distribtued args
-        - Pull from github if required
-        - Launch torchrun command
-        """
-        nodes = get_num_nodes(self.stage_cfg)
-        ntasks_per_node = get_ntasks_per_node(self.stage_cfg)
-        script_text = ["/bin/bash", "set -ex"]
-        # Prepare for the host information to create the torchrun command
-        if nodes > 1:
-            script_text.extend(
-                [
-                    f"{MASTER_ADDR}=$(head -n 1 {str(self._get_hostfile_location())})",
-                    f"{NODEID}=$(head -n 1 {str(self._get_nodeid_location())})",
-                    f"{NNODES}={nodes}",
-                    f"{PROCESSES_PER_NODE}={ntasks_per_node}",
-                    f"{MASTER_PORT}={port}",
-                    "",
-                ]
-            )
-            if self.device == "trainium":
-                script_text.append(
-                    f'{DISTRIBUTED_ARGS}="--nproc_per_node ${PROCESSES_PER_NODE} --nnodes ${NNODES} --node_rank ${NODEID} --master_addr ${MASTER_ADDR} --master_port ${MASTER_PORT}"'
-                )
-            else:
-                script_text.append(
-                    f'{DISTRIBUTED_ARGS}="--nproc_per_node ${PROCESSES_PER_NODE} --nnodes ${NNODES} --rdzv_endpoint=${MASTER_ADDR} --rdzv_id=100 --rdzv_backend=c10d"'
-                )
-        else:
-            script_text = [f'{DISTRIBUTED_ARGS}="--nproc_per_node {ntasks_per_node}"']
-
-        # Prepare github pull
-        # Aligns with the train-script preparition in launcher/nemo/k8s_templates/training.yaml
-        script_text.append("")
-        if self.cfg.get("git", None) is not None or self._default_repo is not None:
-            repo_url = self._default_repo
-            if self.cfg.git.get("repo_url", None) is not None:
-                repo_url = str(self.cfg.git.get("repo_url"))
-            if repo_url is not None:
-                if self.cfg.git.get("token", None) is not None:
-                    repo_url = self.insert_git_token(repo_url, self.cfg.git.token)
-
-                script_text.extend(
-                    [
-                        "mkdir -p $HOME/tmp",
-                        "GIT_CLONE_DIR=$HOME/tmp/$HOSTNAME",
-                        "[[ -d $GIT_CLONE_DIR ]] && rm -rf $GIT_CLONE_DIR",
-                        f"git clone {repo_url} $GIT_CLONE_DIR",
-                        "GIT_CLONE_DIR=${GIT_CLONE_DIR}/",
-                        "cd $GIT_CLONE_DIR",
-                    ]
-                )
-            if self.cfg.git.get("branch", None) is not None:
-                script_text.append(f"git checkout {self.cfg.git.branch}")
-            if self.cfg.git.get("commit", None) is not None:
-                script_text.append(f"git fetch origin {self.cfg.git.commit}")
-                script_text.append(f"git reset --hard {self.cfg.git.commit}")
-        else:
-            script_text.append('GIT_CLONE_DIR=""')
-
-        script_text.append("")
-        script_text.append("unset SLURM_NTASKS")
-
-        script_text.append("")
-        script_text.append(self._make_custom_call_string(stage_cfg_path))
-        return "\n".join(script_text)
-
-    def make_stage_command_groups(self, stage_cfg_path: Path) -> List[List[str]]:
-        """
-        Custom run stage which will invoke the entry script only
-        [TODO] Make this compatiable with NeMo flow as well
-        """
-        # There will be only a single command group
-        command_groups = [[f"bash {stage_cfg_path.parents[0]}/train_script.sh"]]
-        return command_groups
-
-    def run(self) -> str:
-        """
-        Run current stage
-        """
-        # Setup folders and datasets
-        self.setup_folder_and_data()
-        # Save stage hydra config
-        job_path = self.get_job_path()
-
-        is_custom = self.cfg.get("training_cfg") is not None
-        if not is_custom:
-            stage_cfg_path = Training.save_stage_hydra_config(self.stage_cfg, job_path, self.cfg)
-        else:
-            stage_cfg_path = job_path.config_file
-
-        # Make cluster parameters
-        cluster_parameters = self._make_cluster_parameters(self.cluster)
-
-        cluster_parameters["train_script_text"] = self._make_train_script_text(stage_cfg_path)
-        cluster_parameters["hostfile"] = self._get_hostfile_location()
-        cluster_parameters["nodeid"] = self._get_nodeid_location()
-
-        # Make k8s config file if necessary
-        if self.cluster == "k8s":
-            # template_root will be overrided to use local ones
-            template_root = None
-            self._make_k8s_spec_file(template_root, cluster_parameters, job_path, stage_cfg_path)
-            self._copy_k8s_helm_chart(template_root, job_path)
-
-            # k8s does not need command groups
-            command_groups = None
-        else:
-            command_groups = self.make_stage_command_groups(stage_cfg_path)
-        # Create launcher, using SM cutomized launcher
-        launcher = SMAutoLauncher(
-            folder=job_path.folder,
-            cluster=self.cluster,
-            **cluster_parameters,
-        )
-        job_id = launcher.launch(command_groups=command_groups)
-
-        return job_id
-
-    def get_script_args_str(self, stage_cfg_path: Path) -> str:
-        """
-        Based on https://github.com/NVIDIA/NeMo-Framework-Launcher/blob/23.11/launcher_scripts/nemo_launcher/core/stages.py#L608
-        """
-        if self.cluster == "k8s":
-            return "--config-path=/config --config-name=config.yaml"
-        return f"--config-path={stage_cfg_path.parents[0]} --config-name={stage_cfg_path.name}"
-
-    def insert_git_token(self, repo_url: str, token: str) -> str:
-        """
-        Insert git token to git repo url. Currently only support github repo
-        """
-
-        host_name = urlparse(repo_url).hostname
-        if "github.com" == host_name:
-            splitted_url = repo_url.split("github.com", 1)
-            repo_url = splitted_url[0] + self.cfg.git.token + "@github.com" + splitted_url[1]
-        return repo_url
-
-    def _make_nemo_path_command(self) -> List[str]:
-        """Extend nemo path to python path"""
-        # [TODO] clone the nemo/SFA/NxTT repo and handle point to the right path
-        return super()._make_nemo_path_command()
-
-    def _make_container_mounts_string(self) -> str:
-        mounts_string = super()._make_container_mounts_string()
-        # https://github.com/NVIDIA/NeMo-Framework-Launcher/blob/23.11/launcher_scripts/nemo_launcher/core/stages.py#L264
-        # We do not have data dir for custom launching
-        mounts_string = mounts_string.replace(",None:None", "")
-        return mounts_string
-
-    def generate_default_k8s_value_template(self, template_root, cluster_parameters, stage_cfg_path=None):
-        """
-        Setting the general k8s configs that will be applicable for all device types and training methods
-        """
-        with open(os.path.join(template_root, "values.yaml")) as value_file:
-            values_template = OmegaConf.load(value_file)
-        values_template.image.trainingImage = cluster_parameters["container_image"]
-        values_template.trainingConfig.jobName = self.stage_cfg.run.name
-        # Cluster configs
-        if "pullPolicy" in cluster_parameters:
-            values_template.image.pullPolicy = cluster_parameters["pullPolicy"]
-        values_template.trainingConfig.numEFADevices = self.num_efa_devices
-        if "env_vars" in cluster_parameters:
-            values_template.trainingConfig.envVars = cluster_parameters["env_vars"]
-        if "restartPolicy" in cluster_parameters:
-            values_template.trainingConfig.restartPolicy = cluster_parameters["restartPolicy"]
-        if "persistent_volume_claims" in cluster_parameters:
-            values_template.trainingConfig.persistentVolumeClaims = cluster_parameters["persistent_volume_claims"]
-        if "volumes" in cluster_parameters:
-            values_template.trainingConfig.volumes = cluster_parameters["volumes"]
-        if cluster_parameters.get("namespace", None) is not None:
-            values_template.trainingConfig.namespace = cluster_parameters["namespace"]
-        if cluster_parameters.get("annotations", None) is not None:
-            values_template.trainingConfig.annotations = cluster_parameters["annotations"]
-        if cluster_parameters.get("priority_class_name", None) is not None:
-            values_template.trainingConfig.priorityClassName = cluster_parameters["priority_class_name"]
-        if cluster_parameters.get("service_account_name") is not None:
-            values_template.trainingConfig.serviceAccountName = cluster_parameters["service_account_name"]
-        if cluster_parameters.get("custom_labels", None) is not None:
-            values_template.trainingConfig.customLabels = cluster_parameters["custom_labels"]
-        if cluster_parameters.get("label_selector", None) is not None:
-            values_template.trainingConfig.labelSelector = cluster_parameters["label_selector"]
-
-        if self._default_repo is not None:
-            values_template.trainingConfig.git.repo_url = self._default_repo
-
-        # Git configs
-        if self.cfg.get("git", None) is not None:
-            if self.cfg.git.get("repo_url", None) is not None:
-                repo_url = str(self.cfg.git.repo_url)
-                if self.cfg.git.get("token", None) is not None:
-                    repo_url = self.insert_git_token(repo_url, self.cfg.git.token)
-                values_template.trainingConfig.git.repo_url = repo_url
-            if self.cfg.git.get("branch", None) is not None:
-                values_template.trainingConfig.git.branch = self.cfg.git.branch
-            if self.cfg.git.get("commit", None) is not None:
-                values_template.trainingConfig.git.commit = self.cfg.git.commit
-
-        values_template.trainingConfig.device = self.device
-        values_template.trainingConfig.scriptArgs = self.get_script_args_str(stage_cfg_path)
-        return values_template
-
-    def write_value_template(self, values_template, job_path):
-        """
-        Write the value template into disk
-        """
-        k8s_template_path = job_path.folder
-        k8s_template_file = Path(k8s_template_path / "k8s_template" / "values.yaml")
-        k8s_template_file.parent.mkdir(parents=True, exist_ok=True)
-
-        conf = OmegaConf.create(values_template)
-        OmegaConf.save(conf, k8s_template_file)
-
-    def update_stage_specific_k8s_values(self, values_template):
-        """
-        Update the k8s configs that is related to the current stage
-        """
-        values_template.trainingConfig.ntasksPerNode = self.stage_cfg.trainer.devices
-        values_template.trainingConfig.nodes = self.stage_cfg.trainer.num_nodes
-        choice_model_type, _ = self.get_stage_config_choice()
-        if self.cfg.git.get("entry_script", None) is not None:
-            # Override with customer fed entry script
-            values_template.trainingConfig.scriptPath = self.cfg.git.entry_script
-        else:
-            values_template.trainingConfig.scriptPath = str(self._entry_script_path)
-        return values_template
-
-    def _make_k8s_spec_file(
-        self, template_root: str, cluster_parameters: Dict, job_path: JobPaths, stage_cfg_path=None
-    ):
-        """
-        Referring from https://github.com/NVIDIA/NeMo-Framework-Launcher/blob/23.11/launcher_scripts/nemo_launcher/core/stages.py#L669
-        Break the function into 3 parts so we can easily override in different stages
-        - Create general k8s configs that will be applicable for all device types and training methods
-        - Update stage specific k8s configs
-        - Write k8s configs to disk as value.yaml, which will be consumed by helm
-        """
-        #  Need to override the template_root to use our templates
-        # [TODO] Currently hard-code it to do the stage as training
-        template_root = os.path.join(
-            os.path.abspath(os.path.dirname(__file__)),
-            f"k8s_templates/training",
-        )
-        values_template = self.generate_default_k8s_value_template(template_root, cluster_parameters, stage_cfg_path)
-        values_template = self.update_stage_specific_k8s_values(values_template)
-        self.write_value_template(values_template, job_path)
-
-    def _copy_k8s_helm_chart(self, template_root: str, job_path: JobPaths):
-        #  Need to override the template_root to use our templates
-        # [TODO] Currently hard-code it to do the stage as training
-        template_root = os.path.join(
-            os.path.abspath(os.path.dirname(__file__)),
-            f"k8s_templates/training",
-        )
-        super()._copy_k8s_helm_chart(template_root, job_path)
-
-    def get_env_vars(self) -> Dict:
-        """
-        Set up dictionary for environment variables
-        By default injecting the EFA env variable when doing multi-node training
-        The environment variables from hydra config will be set inside the job scripts.
-        For Example:
-            Set `env_vars.NVTE_BIAS_DROPOUT_FUSION=1` while calling nemo_launcherlauncher-scripts,
-            `NVTE_BIAS_DROPOUT_FUSION=1` will be set while running the job.
-
-        :return: a dictionary of env vars while running the job.
-        :rtype: Dict
-        """
-        env_vars = super().get_env_vars()
-        stage_cfg = self.stage_cfg
-        nodes = get_num_nodes(stage_cfg)
-        if int(nodes) > 1:
-            env_vars = set_multinode_envs(env_vars, self.instance_type)
-        return env_vars
-
-
-class SMCustomTraining(SMTraining):
-    """
-    Base stage for the custom training on Sagemaker.
-    """
-
-    @property
-    def _entry_script_path(self) -> Path:
-        return Path(self.stage_cfg.entry_script)
-
-    def setup_stage_vars(self, cfg):
-        """Setup the stage vars, i.e. stage name and stage cfg"""
-        self.stage_name = "custom_training_sm"
-        self.stage_cfg = cfg.get("training_cfg")
-
-    def get_script_args_str(self, stage_cfg_path=None):
-        """
-        Getting all script args and make it as a str
-        """
-        arg_str = []
-        if self.stage_cfg.get("script_args", None) is not None:
-            # script_args will be a list of dict which has key of arg_name and value of arg_value
-            for arg in list(self.stage_cfg.script_args):
-                for key, val in arg.items():
-                    arg_str.append(f"{key} {val} ")
-        return "".join(arg_str)
-
-    def update_stage_specific_k8s_values(self, values_template):
-        """
-        Custom training specifc k8s values
-        """
-        values_template.trainingConfig.ntasksPerNode = get_ntasks_per_node(self.stage_cfg)
-        values_template.trainingConfig.nodes = get_num_nodes(self.stage_cfg)
-        values_template.trainingConfig.scriptPath = self.stage_cfg.entry_script
-        values_template.trainingConfig.customScript = True
-        return values_template
-
-    def _copy_k8s_helm_chart(self, template_root: str, job_path: JobPaths):
-        #  Need to override the template_root to use our templates
-        # [TODO] Currently hard-code it to do the stage as training
-        template_root = os.path.join(
-            os.path.abspath(os.path.dirname(__file__)),
-            f"k8s_templates/training",
-        )
-        # For custom run, there is no need for training config files
-        # Only creating training.yaml, Chart.yaml
-        template_file = os.path.join(template_root, "training.yaml")
-        chart_file = os.path.join(template_root, "Chart.yaml")
-        training_path = Path(job_path.folder / "k8s_template" / "templates" / "training.yaml")
-        training_path.parent.mkdir(parents=True, exist_ok=True)
-        chart_path = Path(job_path.folder / "k8s_template" / "Chart.yaml")
-
-        shutil.copy2(template_file, training_path)
-        shutil.copy2(chart_file, chart_path)
-
-    def _make_cluster_parameters(self, cluster: str) -> Dict:
-        """
-        Make a cluster-specific parameters for jobs on different clusters.
-
-        :param str cluster: i.e. `bcm`, `bcp`, `interactive`, etc.
-        :return: a dictionary of cluster parameters, e.g. `ntasks_per_node`
-        :rtype: Dict
-        """
-        with omegaconf.open_dict(self.cfg):
-            # Patch self.cfg.cluster to align with
-            # https://github.com/NVIDIA/NeMo-Framework-Launcher/blob/23.11/launcher_scripts/nemo_launcher/core/stages.py#L312
-            origin_cluster = self.cfg.cluster
-            self.cfg.cluster = self.cfg.cluster.cluster_config
-            cluster_parameters = super()._make_cluster_parameters(cluster)
-            cluster_type = origin_cluster.get("cluster_type")
-            if cluster_type == "k8s":
-                env_vars = cluster_parameters.get("env_vars")
-                if env_vars and "SLURM_NTASKS_PER_NODE" in env_vars:
-                    env_vars.pop("SLURM_NTASKS_PER_NODE")
-            self.cfg.cluster = origin_cluster
-        return cluster_parameters
-
-
-class SMCustomTrainingGPU(SMCustomTraining):
-    """
-    Stage for training with custom stage on GPU
-    """
-
-    @property
-    def _cuda_visible_devices(self) -> str:
-        ntasks_per_node = get_ntasks_per_node(self.stage_cfg)
-        if ntasks_per_node is None:
-            ntasks_per_node = 8
-        return (
-            "CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7"
-            if ntasks_per_node == 8
-            else f"CUDA_VISIBLE_DEVICES={','.join(map(str, range(ntasks_per_node)))}"
-        )
-
-    @property
-    def _set_ln_sm_margin(self) -> str:
-        return ""
-
-    @property
-    def _skip_ag_overlap(self) -> str:
-        return ""
-
-
-class SMCustomTrainingCPU(SMCustomTrainingGPU):
-    """
-    Stage for custom training on CPU
-    """
-
-    def __init__(self, cfg):
-        super().__init__(cfg)
-        self.device = "cpu"
-
-    @property
-    def _cuda_visible_devices(self) -> str:
-        return ""
-
-
-class SMCustomTrainingTrainium(SMCustomTraining):
-    """
-    Stage for custom training on Trainium
-    """
-
-    def __init__(self, cfg):
-        super().__init__(cfg)
-        self.device = "trainium"
-
-    def make_stage_command_groups(self, stage_cfg_path: Path) -> List[List[str]]:
-        """
-        Make the command groups for current stage
-        Command groups is a list of command group. A command group is defined as:
-              0. Command group is a list of command strings
-              1. Each command group occupies one bcprun, srun or bash
-              2. Each command group eventually has multiple commands connected by ";"
-
-        :param Path stage_cfg_path: path to interpolated and saved configuration
-        :return: command groups for current stage
-        :rtype: List[List[str]]
-        """
-
-    def update_stage_specific_k8s_values(self, values_template):
-        """
-        Custom training specifc k8s values for trainum
-        """
-        super().update_stage_specific_k8s_values(values_template)
-        values_template.trainingConfig.numNeuronDevices = get_num_accelerator_devices(self.instance_type)
-        return values_template
diff --git a/src/hyperpod_cli/custom_launcher/main.py b/src/hyperpod_cli/custom_launcher/main.py
deleted file mode 100755
index cc27626..0000000
--- a/src/hyperpod_cli/custom_launcher/main.py
+++ /dev/null
@@ -1,197 +0,0 @@
-import math
-import os
-import sys
-
-from .validations_wrapper import validate_config
-
-LAUNCHER_SCRIPT_PATH = (
-    f"{os.path.dirname(os.path.abspath(__file__))}/launcher/nemo/nemo_framework_launcher/launcher_scripts/"
-)
-sys.path.append(LAUNCHER_SCRIPT_PATH)
-
-import hydra
-import omegaconf
-from nemo_launcher.core.data_curation_stages import QualityFiltering
-from nemo_launcher.core.data_stages import (
-    CustomDataPreparation,
-    MC4DataPreparation,
-    PileDataPreparation,
-)
-from nemo_launcher.core.export_stages import Export
-from nemo_launcher.core.rlhf_stages import RLHFPPO, RLHFRewardModel
-from nemo_launcher.core.stages import (
-    PEFT,
-    AdapterLearning,
-    Conversion,
-    EvalHarnessEvaluation,
-    FineTuning,
-    IA3Learning,
-    NeMoEvaluation,
-    PromptLearning,
-)
-
-from .launcher.accelerator_devices import (
-    get_num_accelerator_devices,
-    get_num_cores_per_accelerator,
-)
-from .launcher.nemo.stages import (
-    SMCustomTrainingCPU,
-    SMCustomTrainingGPU,
-    SMCustomTrainingTrainium,
-    get_instance_type,
-)
-
-omegaconf.OmegaConf.register_new_resolver("multiply", lambda x, y: x * y, replace=True)
-omegaconf.OmegaConf.register_new_resolver("divide_ceil", lambda x, y: int(math.ceil(x / y)), replace=True)
-omegaconf.OmegaConf.register_new_resolver("divide_floor", lambda x, y: int(math.floor(x / y)), replace=True)
-
-STR2STAGECLASS = {
-    "fine_tuning": FineTuning,
-    "peft": PEFT,
-    "prompt_learning": PromptLearning,
-    "adapter_learning": AdapterLearning,
-    "ia3_learning": IA3Learning,
-    "conversion": Conversion,
-    "export": Export,
-    "evaluation": {
-        EvalHarnessEvaluation: ["gpt3", "prompt_gpt3", "llama", "prompt_llama"],
-        NeMoEvaluation: [
-            "t5",
-            "mt5",
-            "prompt_t5",
-            "prompt_mt5",
-            "adapter_t5",
-            "adapter_gpt3",
-            "ia3_t5",
-            "ia3_gpt3",
-            "peft_llama",
-        ],
-    },
-    "data_preparation": {
-        PileDataPreparation: ["gpt3", "t5", "bert", "llama"],
-        MC4DataPreparation: ["mt5"],
-        CustomDataPreparation: ["generic"],
-    },
-    "rlhf_rm": RLHFRewardModel,
-    "rlhf_ppo": RLHFPPO,
-    "quality_filtering": QualityFiltering,
-}
-
-
-def get_training_stage(cfg):
-    """
-    Get the right training stage based on the device type and if it is custom training
-    """
-    instance_type = get_instance_type(cfg)
-    # p and g instances are GPU instances
-    if instance_type.startswith("p") or instance_type.startswith("g"):
-        device_type = "gpu"
-    elif instance_type.startswith("trn"):
-        device_type = "trainium"
-    else:
-        device_type = "cpu"
-    if cfg.get("training_cfg") is None:
-        raise ValueError("Please provide training_cfg in the YAML file")
-    else:
-        if device_type == "gpu":
-            return SMCustomTrainingGPU
-        if device_type == "trainium":
-            return SMCustomTrainingTrainium
-        return SMCustomTrainingCPU
-
-
-def preprocess_config(cfg):
-    with omegaconf.open_dict(cfg):
-        cfg.launcher_scripts_path = LAUNCHER_SCRIPT_PATH
-    # Override the cluster type to align with NeMo
-    if cfg.get("cluster_type") is None:
-        assert cfg.get("cluster") is not None
-        cluster_type = cfg.cluster.cluster_type
-    else:
-        cluster_type = cfg.cluster_type
-
-    with omegaconf.open_dict(cfg):
-        if cluster_type == "slurm":
-            cfg.cluster_type = "bcm"
-        else:
-            cfg.cluster_type = cluster_type
-
-    if cfg.get("wandb_api_key_file") is None:
-        with omegaconf.open_dict(cfg):
-            cfg.wandb_api_key_file = None
-
-    if cfg.get("wandb_api_bcp_secret_key") is None:
-        with omegaconf.open_dict(cfg):
-            cfg.wandb_api_bcp_secret_key = None
-
-    if cfg.get("training_cfg") is not None:
-        assert cfg.get("stages") is None, "training_cfg and stages should not set together"
-        stage_cfg = cfg.get("training_cfg")
-        assert stage_cfg.get("run") is not None, "run config should be set"
-        run_config = stage_cfg.get("run")
-        if run_config.get("ntasks_per_node") is not None:
-            ntasks_per_node = run_config.get("ntasks_per_node")
-        else:
-            instance_type = get_instance_type(cfg)
-            if instance_type is not None and get_num_accelerator_devices(instance_type) is not None:
-                ntasks_per_node = get_num_accelerator_devices(instance_type) * get_num_cores_per_accelerator(
-                    instance_type
-                )
-            else:
-                ntasks_per_node = 8
-        # To align with https://github.com/NVIDIA/NeMo-Framework-Launcher/blob/23.11/launcher_scripts/nemo_launcher/core/stages.py#L721
-        with omegaconf.open_dict(stage_cfg):
-            stage_cfg.trainer = {"devices": ntasks_per_node}
-            with omegaconf.open_dict(run_config):
-                run_config.ntasks_per_node = ntasks_per_node
-                run_config.results_dir = f"{cfg.base_results_dir}/{run_config.name}"
-        # To align with https://github.com/NVIDIA/NeMo-Framework-Launcher/blob/23.11/launcher_scripts/nemo_launcher/core/stages.py#L313C54-L313C72
-        with omegaconf.open_dict(cfg):
-            cfg.training = {"model": {"ub_tp_comm_overlap": False}}
-        return True
-    return False
-
-
-@hydra.main(config_path="conf", config_name="config", version_base="1.2")
-@validate_config
-def main(cfg):
-    custom_script = preprocess_config(cfg)
-
-    if custom_script:
-        stage_class = get_training_stage(cfg)
-        stage = stage_class(cfg)
-        job_id = stage.run()
-    else:
-        requested_stages = cfg.get("stages")
-
-        dependency = None
-        for stage_name in requested_stages:
-            # Get our training stages
-            if stage_name == "training":
-                stage_class = get_training_stage(cfg)
-            else:
-                stage_class = STR2STAGECLASS[stage_name]
-            if isinstance(stage_class, dict):
-                stage_config_choice = cfg.get(f"{stage_name}_config")
-                choice_model_type = stage_config_choice.rsplit("/", 1)[0]
-                for cls, model_types in stage_class.items():
-                    if choice_model_type in model_types:
-                        stage_class = cls
-                        break
-
-            if dependency is not None:
-                cfg[stage_name]["run"]["dependency"] = dependency
-            stage = stage_class(cfg)
-            job_id = stage.run()
-
-            job_path = stage.get_job_path()
-            command = " \\\n  ".join(sys.argv)
-            with open(job_path.folder / "launcher_cmd.log", "w") as f:
-                f.write(command)
-
-            if job_id:
-                dependency = f"afterany:{job_id}"
-
-
-if __name__ == "__main__":
-    main()
diff --git a/src/hyperpod_cli/custom_launcher/validations_wrapper.py b/src/hyperpod_cli/custom_launcher/validations_wrapper.py
deleted file mode 100644
index 21da7e6..0000000
--- a/src/hyperpod_cli/custom_launcher/validations_wrapper.py
+++ /dev/null
@@ -1,25 +0,0 @@
-from functools import wraps
-from typing import TypeVar, Any, Callable, cast
-
-from omegaconf import DictConfig
-
-from .launcher.config_validator.value_validator import ValueValidator
-from .launcher.config_validator.type_validator import TypeValidator
-
-_T = TypeVar("_T", bound=Callable[..., Any])
-
-
-def validate_config(fn: _T) -> _T:
-    @wraps(fn)
-    def validations_wrapper(config: DictConfig, *args, **kwargs) -> DictConfig:
-        """
-        Execute all validations in this function
-        """
-        type_validator = TypeValidator(config)
-        type_validator.validate()
-        schema_validator = ValueValidator(config)
-        schema_validator.validate()
-
-        return fn(config, *args, **kwargs)
-
-    return cast(_T, validations_wrapper)
diff --git a/src/hyperpod_cli/sagemaker_hyperpod_recipes b/src/hyperpod_cli/sagemaker_hyperpod_recipes
new file mode 160000
index 0000000..000680b
--- /dev/null
+++ b/src/hyperpod_cli/sagemaker_hyperpod_recipes
@@ -0,0 +1 @@
+Subproject commit 000680b2389d58b3a14678ac7ee8dde441228393
diff --git a/src/hyperpod_cli/service/cancel_training_job.py b/src/hyperpod_cli/service/cancel_training_job.py
index dbed1a1..c3f6237 100644
--- a/src/hyperpod_cli/service/cancel_training_job.py
+++ b/src/hyperpod_cli/service/cancel_training_job.py
@@ -18,6 +18,14 @@
 )
 from kubernetes.client.rest import ApiException
 
+from hyperpod_cli.constants.pytorch_constants import (
+    PYTORCH_CUSTOM_OBJECT_GROUP,
+    PYTORCH_CUSTOM_OBJECT_PLURAL
+)
+from hyperpod_cli.service.discover_namespaces import DiscoverNamespaces
+from kubernetes.client import (
+    V1ResourceAttributes
+)
 
 class CancelTrainingJob:
     def __init__(self):
@@ -32,7 +40,14 @@ def cancel_training_job(self, job_name: str, namespace: Optional[str]):
         k8s_client = KubernetesClient()
 
         if not namespace:
-            namespace = k8s_client.get_current_context_namespace()
+            resource_attributes_template = V1ResourceAttributes(
+                verb="delete",
+                group=PYTORCH_CUSTOM_OBJECT_GROUP,
+                resource=PYTORCH_CUSTOM_OBJECT_PLURAL,
+            )
+            namespace = DiscoverNamespaces().discover_accessible_namespace(
+                resource_attributes_template
+            )
         try:
             result = k8s_client.delete_training_job(
                 job_name=job_name, namespace=namespace
diff --git a/src/hyperpod_cli/service/discover_namespaces.py b/src/hyperpod_cli/service/discover_namespaces.py
new file mode 100644
index 0000000..975445e
--- /dev/null
+++ b/src/hyperpod_cli/service/discover_namespaces.py
@@ -0,0 +1,108 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+#     http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+import concurrent
+import sys
+import copy
+from kubernetes.client.rest import ApiException
+from hyperpod_cli.clients.kubernetes_client import KubernetesClient
+from hyperpod_cli.service.get_namespaces import GetNamespaces
+from hyperpod_cli.service.self_subject_access_review import SelfSubjectAccessReview
+from hyperpod_cli.utils import setup_logger
+
+
+logger = setup_logger(__name__)
+
+class DiscoverNamespaces:
+    def __init__(self):
+        return
+
+    def discover_accessible_namespace(self, resource_attributes_template, only_sm_managed=True):
+        """
+        Discover the accessible namespaces
+        """
+        k8s_client = KubernetesClient()
+        context_namespace = k8s_client.get_current_context_namespace()
+
+        # If the namespace is explicitly set by user, take the namespace from the config instead
+        # of discovering automatically 
+        if context_namespace is not None:
+            return context_namespace
+        
+        try:
+            if only_sm_managed:
+                namespaces = GetNamespaces().get_sagemaker_managed_namespaces()
+            else:
+                namespaces = GetNamespaces().get_namespaces()
+
+            discovered_namespaces = self.get_namespaces_by_checking_access_permission(
+                namespaces,
+                resource_attributes_template,
+            )
+
+            if len(discovered_namespaces) == 0:
+                logger.error("Found no accessible namespaces. Please ask for cluster admin for assistance or specify value of namespace explicitly in the command.")
+                sys.exit(1)
+            if len(discovered_namespaces) > 1:
+                logger.error(f"Found more than 1 accessible namespaces {discovered_namespaces}. Please specify value of namespace explicitly in the command.")
+                sys.exit(1)
+            
+            if len(discovered_namespaces) == 1:
+                logger.info(f"Found accessible namespace: {discovered_namespaces}")
+            return discovered_namespaces[0]
+        except ApiException as e:
+            if e.status == 403:
+                logger.error("Got access denied error when discovering accessible namespaces, please specify the '--namespace' parameter in the command and try again.")
+                sys.exit(1)
+
+
+    def get_namespaces_by_checking_access_permission(
+            self,
+            namespaces,
+            resource_attributes_template, 
+            max_workers=10
+        ):
+        """
+        Get the accessible namespaces by performing the SelfSubjectAccessReview. For each of the namespace,
+        check if user has specified permission to it, if the answer is NO, the namespace will be skipped.
+
+        Performing access check can take quite long if the number of namespaces is large. Thus the implementation
+        leverages the multi-threading to ensure that multiple access check can be performed in parallel.
+        """
+        subject_access_review = SelfSubjectAccessReview()
+        accessible_namespaces = list()
+        resource_attributes = list()
+
+        for namespace in namespaces:
+            resource_attribute = copy.deepcopy(resource_attributes_template)
+            resource_attribute.namespace = namespace
+            resource_attributes.append(resource_attribute)
+
+        # Multi-thread the self subject access review to improve the performance
+        with concurrent.futures.ThreadPoolExecutor(max_workers=max_workers) as executor:
+            futures = {
+                executor.submit(
+                    subject_access_review.self_subject_access_review, resource_attribute
+                ) for resource_attribute in resource_attributes
+            }
+            
+            for future in concurrent.futures.as_completed(futures):
+                try:
+                    response = future.result()
+                    if response.status.allowed:
+                        accessible_namespaces.append(
+                            response.spec.resource_attributes.namespace
+                        )
+                except Exception as e:
+                    raise(e)
+
+        return accessible_namespaces
\ No newline at end of file
diff --git a/src/hyperpod_cli/service/get_logs.py b/src/hyperpod_cli/service/get_logs.py
index ac82a53..7dd381a 100644
--- a/src/hyperpod_cli/service/get_logs.py
+++ b/src/hyperpod_cli/service/get_logs.py
@@ -15,11 +15,12 @@
 from hyperpod_cli.clients.kubernetes_client import (
     KubernetesClient,
 )
+from hyperpod_cli.service.discover_namespaces import DiscoverNamespaces
 from hyperpod_cli.service.list_pods import (
     ListPods,
 )
 from kubernetes.client.rest import ApiException
-
+from kubernetes.client import V1ResourceAttributes
 
 class GetLogs:
     def __init__(self):
@@ -38,7 +39,15 @@ def get_training_job_logs(
         list_pods_service = ListPods()
 
         if not namespace:
-            namespace = k8s_client.get_current_context_namespace()
+            resource_attributes_template = V1ResourceAttributes(
+                verb="get",
+                group="",
+                resource="pods",
+                subresource="log",
+            )
+            namespace = DiscoverNamespaces().discover_accessible_namespace(
+                resource_attributes_template
+            )
 
         try:
             pods_for_training_job = list_pods_service.list_pods_for_training_job(
diff --git a/src/hyperpod_cli/service/get_namespaces.py b/src/hyperpod_cli/service/get_namespaces.py
new file mode 100644
index 0000000..ecb39ce
--- /dev/null
+++ b/src/hyperpod_cli/service/get_namespaces.py
@@ -0,0 +1,60 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+#     http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+from hyperpod_cli.clients.kubernetes_client import (
+    KubernetesClient,
+)
+from kubernetes.client.rest import ApiException
+
+from hyperpod_cli.constants.command_constants import SAGEMAKER_MANAGED_QUEUE_LABEL
+
+
+LIMIT_PER_REQUEST = 200
+
+class GetNamespaces:
+
+    def __init__(self):
+        return
+
+    def get_namespaces(self, label_selector=None):
+        """
+        Get namespaces in the cluster
+        """
+        core_v1_api = KubernetesClient().get_core_v1_api()
+        all_namespaces = list()
+        continue_token = None
+
+        while True:
+            response = None
+            if continue_token:
+                response = core_v1_api.list_namespace(
+                    limit=LIMIT_PER_REQUEST, label_selector=label_selector, _continue=continue_token
+                )
+            else:
+                response = core_v1_api.list_namespace(
+                    limit=LIMIT_PER_REQUEST, label_selector=label_selector
+                )
+
+            all_namespaces.extend([ns.metadata.name for ns in response.items])
+            continue_token = response.metadata._continue
+
+            if not continue_token:
+                break
+        
+        return all_namespaces
+
+    def get_sagemaker_managed_namespaces(self):
+        """
+        Get sagemaker managed namespaces in the cluster
+        """
+
+        return self.get_namespaces(SAGEMAKER_MANAGED_QUEUE_LABEL + "=true")
diff --git a/src/hyperpod_cli/service/get_training_job.py b/src/hyperpod_cli/service/get_training_job.py
index 685f696..fb12147 100644
--- a/src/hyperpod_cli/service/get_training_job.py
+++ b/src/hyperpod_cli/service/get_training_job.py
@@ -20,7 +20,12 @@
 
 from hyperpod_cli import utils
 from kubernetes.client.rest import ApiException
+from kubernetes.client import (
+    V1ResourceAttributes
+)
 
+from hyperpod_cli.constants.pytorch_constants import PYTORCH_CUSTOM_OBJECT_GROUP, PYTORCH_CUSTOM_OBJECT_PLURAL
+from hyperpod_cli.service.discover_namespaces import DiscoverNamespaces
 
 class GetTrainingJob:
     def __init__(self):
@@ -33,14 +38,21 @@ def get_training_job(
         verbose: Optional[bool],
     ):
         """
-        Describe training job provided by the user in the speified namespace.
+        Describe training job provided by the user in the specified namespace.
         If namespace is not provided job is described from the default namespace in user context
         """
 
         k8s_client = KubernetesClient()
-
         if not namespace:
-            namespace = k8s_client.get_current_context_namespace()
+            resource_attributes_template = V1ResourceAttributes(
+                verb="get",
+                group=PYTORCH_CUSTOM_OBJECT_GROUP,
+                resource=PYTORCH_CUSTOM_OBJECT_PLURAL,
+            )
+            namespace = DiscoverNamespaces().discover_accessible_namespace(
+                resource_attributes_template
+            )
+
         try:
             result = k8s_client.get_job(
                 job_name=job_name,
diff --git a/src/hyperpod_cli/service/list_pods.py b/src/hyperpod_cli/service/list_pods.py
index 7a082c7..9798611 100644
--- a/src/hyperpod_cli/service/list_pods.py
+++ b/src/hyperpod_cli/service/list_pods.py
@@ -20,7 +20,12 @@
     KubernetesClient,
 )
 from kubernetes.client.rest import ApiException
+from kubernetes.client import (
+    V1ResourceAttributes
+)
 
+from hyperpod_cli.constants.command_constants import NVIDIA_GPU_RESOURCE_LIMIT_KEY
+from hyperpod_cli.service.discover_namespaces import DiscoverNamespaces
 
 class ListPods:
     def __init__(self):
@@ -38,7 +43,14 @@ def list_pods_for_training_job(
         k8s_client = KubernetesClient()
 
         if not namespace:
-            namespace = k8s_client.get_current_context_namespace()
+            resource_attributes_template = V1ResourceAttributes(
+                verb="list",
+                group="",
+                resource="pods",
+            )
+            namespace = DiscoverNamespaces().discover_accessible_namespace(
+                resource_attributes_template
+            )
 
         label_filter = f"training.kubeflow.org/job-name={job_name}"
 
@@ -75,7 +87,7 @@ def list_pods_and_get_requested_resources_group_by_node_name(
             if node_name:
                 for container in pod.spec.containers:
                     if container.resources and container.resources.requests:
-                        gpu_request = container.resources.requests.get("nvidia.com/gpu")
+                        gpu_request = container.resources.requests.get(NVIDIA_GPU_RESOURCE_LIMIT_KEY)
                         neuron_request = container.resources.requests.get(
                             "aws.amazon.com/neurondevice"
                         )
diff --git a/src/hyperpod_cli/service/list_training_jobs.py b/src/hyperpod_cli/service/list_training_jobs.py
index 2aa1aa2..422a1b4 100644
--- a/src/hyperpod_cli/service/list_training_jobs.py
+++ b/src/hyperpod_cli/service/list_training_jobs.py
@@ -19,6 +19,13 @@
     KubernetesClient,
 )
 from kubernetes.client.rest import ApiException
+from kubernetes.client import (
+    V1ResourceAttributes
+)
+
+from hyperpod_cli.constants.command_constants import KUEUE_WORKLOAD_PRIORITY_CLASS_LABEL_KEY
+from hyperpod_cli.constants.pytorch_constants import PYTORCH_CUSTOM_OBJECT_GROUP, PYTORCH_CUSTOM_OBJECT_PLURAL
+from hyperpod_cli.service.discover_namespaces import DiscoverNamespaces
 
 
 class ListTrainingJobs:
@@ -54,7 +61,14 @@ def list_training_jobs(
                             jobs.append(_job)
             else:
                 if not namespace:
-                    namespace = k8s_client.get_current_context_namespace()
+                    resource_attributes_template = V1ResourceAttributes(
+                        verb="list",
+                        group=PYTORCH_CUSTOM_OBJECT_GROUP,
+                        resource=PYTORCH_CUSTOM_OBJECT_PLURAL,
+                    )
+                    namespace = DiscoverNamespaces().discover_accessible_namespace(
+                        resource_attributes_template
+                    )
 
                 namespace_jobs = k8s_client.list_training_jobs(
                     namespace=namespace,
@@ -76,17 +90,22 @@ def _generate_list_training_job_output(self, jobs: List):
                 namespace = job.get("metadata").get("namespace")
                 creation_time = None
                 state = None
+                priority = self._get_job_priority(job)
                 if job.get("status"):
                     creation_time = job.get("status").get("startTime")
                     state = self._get_job_status(job.get("status").get("conditions"))
-                output_jobs["jobs"].append(
-                    {
-                        "Name": name,
-                        "Namespace": namespace,
-                        "CreationTime": creation_time,
-                        "State": state,
-                    }
-                )
+
+                job_summary = {
+                    "Name": name,
+                    "Namespace": namespace,
+                    "CreationTime": creation_time,
+                    "State": state,
+                }
+
+                if priority is not None:
+                    job_summary["priority"] = priority
+
+                output_jobs["jobs"].append(job_summary)
 
         return json.dumps(output_jobs, indent=1, sort_keys=False)
 
@@ -101,7 +120,26 @@ def _get_job_status(self, status: List) -> Optional[str]:
                 state.get("lastTransitionTime"),
                 "%Y-%m-%dT%H:%M:%SZ",
             )
-            if state_date_time > last_date_time:
+            if state_date_time >= last_date_time:
                 last_date_time = state_date_time
                 current_status = state.get("type")
         return current_status
+
+    def _get_job_priority(self, job) -> Optional[str]:
+        worker_template = job.get("spec", {}).get("pytorchReplicaSpecs", {}).get("Worker", {}).get("template", {})
+        wl_priority_cls = worker_template.get("metadata", {}).get("labels", {}).get(KUEUE_WORKLOAD_PRIORITY_CLASS_LABEL_KEY, None)
+        spec_priority_cls = worker_template.get("spec", {}).get("priorityClassName", None)
+
+        # For reference: https://kueue.sigs.k8s.io/docs/concepts/workload_priority_class/
+        # The workload priority class takes precedence over the k8s priority class.
+        # Because the cli focuses on the job level which means workload priority is more essential.
+        # There is possibility that these two priorities coexist at the same time. In this case,
+        # the k8s priority class will be used as pod priority. cli should still take workload
+        # priority in this scenario.
+
+        if wl_priority_cls is not None:
+            return wl_priority_cls
+        elif spec_priority_cls is not None:
+            return spec_priority_cls
+        
+        return None
diff --git a/src/hyperpod_cli/service/self_subject_access_review.py b/src/hyperpod_cli/service/self_subject_access_review.py
new file mode 100644
index 0000000..dcb326f
--- /dev/null
+++ b/src/hyperpod_cli/service/self_subject_access_review.py
@@ -0,0 +1,48 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+#     http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+from hyperpod_cli.clients.kubernetes_client import (
+    KubernetesClient,
+)
+from kubernetes.client.rest import ApiException
+from kubernetes.client import (
+    V1SelfSubjectAccessReview, 
+    V1SelfSubjectAccessReviewSpec, 
+)
+
+
+class SelfSubjectAccessReview:
+    def __init__(self):
+        return
+
+    def self_subject_access_review(
+            self, 
+            resource_attributes=None, 
+            non_resource_attributes=None, 
+            local_vars_configuration=None,
+        ):
+        """
+        Submit self subject access review
+        """
+        auth_v1_api = KubernetesClient().get_auth_v1_api()
+
+        access_review = V1SelfSubjectAccessReview(
+            spec=V1SelfSubjectAccessReviewSpec(
+                resource_attributes=resource_attributes,
+                non_resource_attributes=non_resource_attributes,
+                local_vars_configuration=local_vars_configuration,
+            )
+        )
+
+        response = auth_v1_api.create_self_subject_access_review(body=access_review)
+        
+        return response
diff --git a/src/hyperpod_cli/validators/job_validator.py b/src/hyperpod_cli/validators/job_validator.py
index 6e817a1..fbdb5b9 100644
--- a/src/hyperpod_cli/validators/job_validator.py
+++ b/src/hyperpod_cli/validators/job_validator.py
@@ -14,13 +14,17 @@
 import os
 import yaml
 from yaml.loader import SafeLoader
-from typing import Optional
+from typing import Optional, List
 
+from hyperpod_cli.clients.kubernetes_client import KubernetesClient
 from hyperpod_cli.constants.command_constants import (
+    KUEUE_WORKLOAD_PRIORITY_CLASS_LABEL_KEY,
+    SAGEMAKER_TRAINING_LAUNCHER_DIR,
     RestartPolicy,
     KUEUE_QUEUE_NAME_LABEL_KEY,
     HYPERPOD_AUTO_RESUME_ANNOTATION_KEY,
     HYPERPOD_MAX_RETRY_ANNOTATION_KEY,
+    SchedulerType
 )
 from hyperpod_cli.constants.hyperpod_instance_types import (
     HyperpodInstanceType,
@@ -32,6 +36,7 @@
 
 logger = setup_logger(__name__)
 
+RECIPES_DIR = os.path.join(SAGEMAKER_TRAINING_LAUNCHER_DIR, "recipes_collection/recipes")
 
 class JobValidator(Validator):
     def __init__(self):
@@ -55,6 +60,7 @@ def validate_start_job_args(
         max_retry: Optional[int],
         namespace: Optional[str],
         entry_script: Optional[str],
+        recipe: Optional[str] = None,
     ):
         if job_kind is not None and job_kind != "kubeflow/PyTorchJob":
             logger.error("The only supported 'job-kind' is 'kubeflow/PyTorchJob'.")
@@ -64,19 +70,25 @@ def validate_start_job_args(
             logger.error("The only supported 'command' is 'torchrun'.")
             return False
 
-        if scheduler_type is not None and scheduler_type != "Kueue":
-            logger.error("The only supported 'scheduler_type' is 'Kueue'.")
+        if scheduler_type is not None and scheduler_type not in SchedulerType.get_values():
+            logger.error(f"The only supported 'scheduler_type' are {SchedulerType.get_values()}.")
             return False
 
         if config_file is not None and job_name is not None:
             logger.error(
-                "Please provide only 'config-file' to submit job using config file or 'job-name' to submit job via CLI arguments"
+                "Please provide only 'config-file' to submit job using custom script or 'job-name' to submit job via CLI arguments"
+            )
+            return False
+        
+        if config_file is not None and recipe is not None:
+            logger.error(
+                "Please provide only 'config-file' to submit job using custom script or 'recipe' for recipe-based jobs"
             )
             return False
 
-        if config_file is None and job_name is None:
+        if config_file is None and job_name is None and recipe is None:
             logger.error(
-                "Please provide either 'config-file' to submit job using config file or 'job-name' to submit job via CLI arguments"
+                "Please provide either 'recipe' for recipe-based jobs or 'config-file' to submit job using config file or 'job-name' to submit job via CLI arguments"
             )
             return False
 
@@ -106,6 +118,10 @@ def validate_start_job_args(
                         "Please ensure 'label-selector' keys are string type and values are string or list of string type"
                     )
                     return False
+            
+            if not validate_scheduler_related_fields(scheduler_type, namespace, priority):
+                return False
+            
             return validate_hyperpod_related_fields(
                 instance_type,
                 queue_name,
@@ -113,23 +129,18 @@ def validate_start_job_args(
                 auto_resume,
                 restart_policy,
                 max_retry,
-                namespace,
+                namespace
             )
+        if recipe is not None:
+            return validate_recipe_file(recipe)
 
         return True
 
-    def validate_start_job_config_yaml(self, file):
-        if not os.path.exists(file):
-            logger.error(f"Configuration file {file} does not exist.")
-            return False
-        config_data = verify_and_load_yaml(file)
-        if config_data is None:
-            return False
-
-        return validate_yaml_content(config_data)
-
 
 def verify_and_load_yaml(file_path: str):
+    if not os.path.exists(file_path):
+        logger.error(f"Configuration file {file_path} does not exist.")
+        return None
     try:
         with open(file_path, "r") as file:
             # Attempt to load the YAML file
@@ -161,7 +172,13 @@ def validate_yaml_content(data):
     custom_labels = cluster_config_fields.get("custom_labels")
     annotations = cluster_config_fields.get("annotations")
     namespace = cluster_config_fields.get("namespace")
+    scheduler_type = cluster_config_fields.get("scheduler_type", SchedulerType.get_default().value)
 
+    if scheduler_type not in SchedulerType.get_values():
+        logger.error(
+            f"Unsupported scheduler type '{scheduler_type}', only {SchedulerType.get_values()} are allowed."
+        )
+        return False
     instance_type = cluster_fields.get("instance_type", None)
     queue_name = None
     if custom_labels is not None:
@@ -182,8 +199,15 @@ def validate_yaml_content(data):
 
     priority = cluster_config_fields.get("priority_class_name", None)
     restart_policy = cluster_config_fields.get("restartPolicy", None)
+    workload_priority = None
+
+    if custom_labels is not None:
+        workload_priority = custom_labels.get(KUEUE_WORKLOAD_PRIORITY_CLASS_LABEL_KEY, None)
+
+    if not validate_scheduler_related_fields(scheduler_type, namespace, workload_priority):
+        return False
 
-    return validate_hyperpod_related_fields(
+    if not validate_hyperpod_related_fields(
         instance_type,
         queue_name,
         priority,
@@ -191,7 +215,10 @@ def validate_yaml_content(data):
         restart_policy,
         max_retry,
         namespace,
-    )
+    ):
+        return False
+
+    return True
 
 
 def validate_hyperpod_related_fields(
@@ -201,8 +228,9 @@ def validate_hyperpod_related_fields(
     auto_resume: bool,
     restart_policy: Optional[RestartPolicy],
     max_retry: Optional[int],
-    namespace: Optional[str],
+    namespace: Optional[str]
 ):
+    logger.info(f"instance_type: {instance_type}, queue_name: {queue_name}, priority: {priority}, auto_resume: {auto_resume}, restart_policy: {restart_policy}, max_retry: {max_retry}, namespace: {namespace}")
     if instance_type is None:
         logger.error(
             "Please provide 'instance-type' to specify instance type for training job"
@@ -223,14 +251,42 @@ def validate_hyperpod_related_fields(
         )
         return False
 
-    if (queue_name is None and priority is not None) or (
-        queue_name is not None and priority is None
-    ):
-        logger.error("Must provide both or neither of 'queue_name' and 'priority'.")
-        return False
     return True
 
 
+def validate_scheduler_related_fields(
+    scheduler_type: SchedulerType,
+    namespace: Optional[str],
+    priority: Optional[str],
+): 
+    if scheduler_type == SchedulerType.SAGEMAKER.value:
+        k8s_client = KubernetesClient()
+        sm_managed_namespace = k8s_client.get_sagemaker_managed_namespace(namespace)
+        if namespace and not sm_managed_namespace:
+            logger.error(
+                f"Scheduler type is '{SchedulerType.SAGEMAKER.value}' however cannot find namespace '{namespace}' managed by SageMaker. Please ensure namespace exists and you have 'get' access to it."
+            )
+            return False
+
+        if priority is not None :       
+            priority_classes = [item['metadata']['name'] for item in k8s_client.list_workload_priority_classes()['items']]
+            if priority not in priority_classes:
+                logger.error(
+                    f"Workload priority class '{priority}' is not found. Please ensure the priority exists and you have 'get' access to 'WorkloadPriorityClass'. Valid priority values are {priority_classes}."
+                )
+                return False
+    return True
+
+def validate_recipe_file(recipe: str):
+    full_recipe_path = os.path.join(RECIPES_DIR, f"{recipe}.yaml")
+    
+    if os.path.exists(full_recipe_path) and os.path.isfile(full_recipe_path):
+        logger.info(f"Recipe file found: {full_recipe_path}")
+        return True
+    
+    logger.error(f"Recipe file '{recipe}.yaml' not found in {RECIPES_DIR}")
+    return False
+
 def is_dict_str_list_str(data: dict) -> bool:
     """
     Check if the given dictionary is of type Dict[str, List[str]].
@@ -249,8 +305,6 @@ def is_dict_str_list_str(data: dict) -> bool:
         ):
             return False
     return True
-
-
 def _validate_json_str(
     json_str: str,
 ):
@@ -274,4 +328,4 @@ def _validate_json_str(
     except Exception as e:
         # Catch any other exceptions
         logger.error(f"An unexpected error occurred: {e}")
-        return False
+        return False
\ No newline at end of file
diff --git a/test/integration_tests/abstract_integration_tests.py b/test/integration_tests/abstract_integration_tests.py
index fd64acd..f881d40 100644
--- a/test/integration_tests/abstract_integration_tests.py
+++ b/test/integration_tests/abstract_integration_tests.py
@@ -19,6 +19,8 @@
 from botocore.exceptions import ClientError
 
 from hyperpod_cli.utils import setup_logger
+from kubernetes.client.rest import ApiException
+from kubernetes import client, config
 
 logger = setup_logger(__name__)
 
@@ -36,12 +38,13 @@ class AbstractIntegrationTests:
     vpc_stack_name = "hyperpod-cli-vpc-stack"
     eks_cluster_name = "hyperpod-cli-cluster-" + suffix
     bucket_name = "hyperpod-cli-s3-" + suffix
+    test_team_name = "test-team"
 
     def _create_session(self):
         session = boto3.Session()
         return session
 
-    def create_test_resorces(self, session):
+    def create_test_resources(self, session):
         cfn = session.client("cloudformation")
 
         # Get static resources from static-resource stack
@@ -169,7 +172,7 @@ def delete_hyperpod_cluster(self, session):
         sagemaker_client.delete_cluster(ClusterName=self.hyperpod_cli_cluster_name)
 
         time.sleep(10)
-        # Wait for sagemkaer stack to create complete
+        # Wait for sagemaker stack to create complete
         try:
             result = self.get_hyperpod_cluster_status(sagemaker_client)
             while result.get("ClusterStatus") == "Deleting":
@@ -239,7 +242,7 @@ def apply_helm_charts(self):
         except subprocess.CalledProcessError as e:
             raise RuntimeError(f"Failed to update helm charts: {e}")
 
-        apply_commanmd = [
+        apply_command = [
             "helm",
             "install",
             "dependencies",
@@ -252,7 +255,7 @@ def apply_helm_charts(self):
             # Execute the command to apply helm charts
             logger.info(
                 subprocess.run(
-                    apply_commanmd,
+                    apply_command,
                     check=True,
                     capture_output=True,
                     text=True,
@@ -261,12 +264,158 @@ def apply_helm_charts(self):
         except subprocess.CalledProcessError as e:
             raise RuntimeError(f"Failed to apply helm charts: {e}")
 
+    def install_kueue(self):
+        command = ["./helm_chart/install_dependencies.sh"]
+        wait_command = ["kubectl", "wait", "deploy/kueue-controller-manager", "-nkueue-system", "--for=condition=available", "--timeout=5m"]
+        try:
+            # Execute the dependencies installation script to install kueue
+            logger.info(
+                subprocess.run(
+                    command,
+                    check=True,
+                    capture_output=True,
+                    text=True,
+                )
+            )
+
+            # Wait for kueue to be available
+            logger.info(
+                subprocess.run(
+                    wait_command,
+                    check=True,
+                    capture_output=True,
+                    text=True,
+                )
+            )
+        except subprocess.CalledProcessError as e:
+            raise RuntimeError(f"Failed to install the dependencies: {e}")
+
+    # TODO: Manually setup quota allocation for now. Migrate to sagemaker public APIs afterwards
+    def create_quota_allocation_resources(self):
+        config.load_kube_config()
+        # Create an instance of the API class
+        core_api = client.CoreV1Api()
+        custom_api = client.CustomObjectsApi()
+
+        try:
+            # Setup namespace 
+            namespace = client.V1Namespace(
+                metadata=client.V1ObjectMeta(
+                    name=f"hyperpod-ns-{self.test_team_name}",
+                    labels={
+                        "sagemaker.amazonaws.com/sagemaker-managed-queue": "true",
+                        "sagemaker.amazonaws.com/quota-allocation-id": self.test_team_name,
+                    }
+                )
+            )
+            core_api.create_namespace(body=namespace)
+            logger.info("Namespace created successfully")
+        except ApiException as e:
+            if e.status == 409:
+                logger.info("Already exists, move on")
+            else:
+                raise e
+        
+        try:
+            # Setup resource flavor
+            resource_flavor = {
+                "apiVersion": "kueue.x-k8s.io/v1beta1",
+                "kind": "ResourceFlavor",
+                "metadata": {
+                    "name": "ml.c5.2xlarge"
+                }
+            }
+            custom_api.create_cluster_custom_object(
+                group="kueue.x-k8s.io",
+                version="v1beta1",
+                plural="resourceflavors",
+                body=resource_flavor
+            )
+            logger.info("ResourceFlavor created successfully")
+        except ApiException as e:
+            if e.status == 409:
+                logger.info("Already exists, move on")
+            else:
+                raise e
+        
+        try:
+            # Setup cluster queue
+            cluster_queue = {
+                "apiVersion": "kueue.x-k8s.io/v1beta1",
+                "kind": "ClusterQueue",
+                "metadata": {
+                    "name": f"hyperpod-ns-{self.test_team_name}-clusterqueue"
+                },
+                "spec": {
+                    "resourceGroups": [
+                        {
+                            "coveredResources": ["cpu", "memory"],
+                            "flavors": [
+                                {
+                                    "name": "ml.c5.2xlarge",
+                                    "resources": [
+                                        {
+                                            "name": "cpu",
+                                            "nominalQuota": 2
+                                        },
+                                        {
+                                            "name": "memory",
+                                            "nominalQuota": "2Gi"
+                                        }
+                                    ]
+                                }
+                            ]
+                        }
+                    ]
+                }
+            }
+            custom_api.create_cluster_custom_object(
+                group="kueue.x-k8s.io",
+                version="v1beta1",
+                plural="clusterqueues",
+                body=cluster_queue
+            )
+            logger.info("ClusterQueue created successfully")
+        except ApiException as e:
+            if e.status == 409:
+                logger.info("Already exists, move on")
+            else:
+                raise e
+
+        try:
+            # Setup local queue
+            local_queue = {
+                "apiVersion": "kueue.x-k8s.io/v1beta1",
+                "kind": "LocalQueue",
+                "metadata": {
+                    "name": f"hyperpod-ns-{self.test_team_name}-localqueue",
+                    "namespace": f"hyperpod-ns-{self.test_team_name}"
+                },
+                "spec": {
+                    "clusterQueue": f"hyperpod-ns-{self.test_team_name}-clusterqueue"
+                }
+            }
+            custom_api.create_namespaced_custom_object(
+                group="kueue.x-k8s.io",
+                version="v1beta1",
+                namespace=f"hyperpod-ns-{self.test_team_name}",
+                plural="localqueues",
+                body=local_queue
+            )
+        except ApiException as e:
+            if e.status == 409:
+                logger.info("Already exists, move on")
+            else:
+                raise e
+
     def setup(self):
         self.new_session = self._create_session()
-        self.create_test_resorces(self.new_session)
+        self.create_test_resources(self.new_session)
         self.create_kube_context()
         self.apply_helm_charts()
         self.create_hyperpod_cluster(self.new_session)
+        self.install_kueue()
+        self.create_quota_allocation_resources()
 
     def tearDown(self):
         self.delete_hyperpod_cluster(self.new_session)
diff --git a/test/integration_tests/data/basicJob.yaml b/test/integration_tests/data/basicJob.yaml
index 5b0b406..daa1f49 100644
--- a/test/integration_tests/data/basicJob.yaml
+++ b/test/integration_tests/data/basicJob.yaml
@@ -23,8 +23,8 @@ training_cfg:
  script_args: []
  run:
   name: hyperpod-cli-test # Current run name
-  nodes: 2 # Number of nodes to use for current training
-  # ntasks_per_node: 1 # Number of devices to use per node
+  nodes: 1 # Number of nodes to use for current training
+  ntasks_per_node: 1 # Number of devices to use per node
 cluster:
  cluster_type: k8s # currently k8s only
  instance_type: ml.c5.2xlarge
@@ -33,7 +33,7 @@ cluster:
   service_account_name: null
   # persistent volume, usually used to mount FSx
   persistent_volume_claims: null
-  namespace: kubeflow
+  namespace: hyperpod-ns-test-team
   # required node affinity to select nodes with HyperPod
   # labels and passed health check if burn-in enabled
   label_selector:
@@ -47,6 +47,7 @@ cluster:
           - 100
   pullPolicy: IfNotPresent # policy to pull container, can be Always, IfNotPresent and Never
   restartPolicy: OnFailure # restart policy
+  scheduler_type: Kueue
 
 base_results_dir: ./result # Location to store the results, checkpoints and logs.
 container: docker.io/kubeflowkatib/pytorch-mnist-cpu:v1beta1-bc09cfd # container to use
diff --git a/test/integration_tests/data/basicJobWithQuota.yaml b/test/integration_tests/data/basicJobWithQuota.yaml
new file mode 100644
index 0000000..0422592
--- /dev/null
+++ b/test/integration_tests/data/basicJobWithQuota.yaml
@@ -0,0 +1,54 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+#     http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+defaults:
+ - override hydra/job_logging: stdout
+
+hydra:
+ run:
+  dir: .
+ output_subdir: null
+
+training_cfg:
+ entry_script: /opt/pytorch-mnist/mnist.py
+ script_args: []
+ run:
+  name: hyperpod-cli-test-with-quota # Current run name
+  nodes: 1 # Number of nodes to use for current training
+  ntasks_per_node: 1 # Number of devices to use per node
+cluster:
+ cluster_type: k8s # currently k8s only
+ instance_type: ml.c5.2xlarge
+ cluster_config:
+  # name of service account associated with the namespace
+  service_account_name: null
+  # persistent volume, usually used to mount FSx
+  persistent_volume_claims: null
+  # required node affinity to select nodes with HyperPod
+  # labels and passed health check if burn-in enabled
+  label_selector:
+      required:
+          sagemaker.amazonaws.com/node-health-status:
+              - Schedulable
+      preferred:
+          sagemaker.amazonaws.com/deep-health-check-status:
+              - Passed
+      weights:
+          - 100
+  pullPolicy: IfNotPresent # policy to pull container, can be Always, IfNotPresent and Never
+  restartPolicy: OnFailure # restart policy
+  scheduler_type: SageMaker
+base_results_dir: ./result # Location to store the results, checkpoints and logs.
+container: docker.io/kubeflowkatib/pytorch-mnist-cpu:v1beta1-bc09cfd # container to use
+
+env_vars:
+ NCCL_DEBUG: INFO # Logging level for NCCL. Set to "INFO" for debug information
\ No newline at end of file
diff --git a/test/integration_tests/test_happy_case.py b/test/integration_tests/test_happy_case.py
index 56bc030..90eae71 100644
--- a/test/integration_tests/test_happy_case.py
+++ b/test/integration_tests/test_happy_case.py
@@ -25,7 +25,6 @@
 
 
 class TestHappyCase(AbstractIntegrationTests):
-    namespace = "kubeflow"
 
     @pytest.fixture(scope="class", autouse=True)
     def basic_test(self):
@@ -54,8 +53,6 @@ def test_hyperpod_connect_cluster(self):
             super().hyperpod_cli_cluster_name,
             "--region",
             "us-west-2",
-            "--namespace",
-            self.namespace,
         ]
 
         result = self._execute_test_command(command)
@@ -78,6 +75,22 @@ def test_start_job(self):
         assert result.returncode == 0
         logger.info(result.stdout)
 
+    @pytest.mark.order(2)
+    def test_start_job_with_quota(self):
+        config_path = os.path.expanduser("./test/integration_tests/data/basicJobWithQuota.yaml")
+        command = [
+            "hyperpod",
+            "start-job",
+            "--config-file",
+            config_path,
+        ]
+
+        result = self._execute_test_command(command)
+        # wait for job to complete creation
+        time.sleep(240)
+        assert result.returncode == 0
+        logger.info(result.stdout)
+
     @pytest.mark.order(3)
     def test_get_job(self):
         command = [
@@ -91,6 +104,20 @@ def test_get_job(self):
         assert result.returncode == 0
         assert ("hyperpod-cli-test") in str(result.stdout)
         logger.info(result.stdout)
+    
+    @pytest.mark.order(3)
+    def test_get_job_with_quota(self):
+        command = [
+            "hyperpod",
+            "get-job",
+            "--job-name",
+            "hyperpod-cli-test-with-quota",
+        ]
+
+        result = self._execute_test_command(command)
+        assert result.returncode == 0
+        assert ("hyperpod-cli-test-with-quota") in str(result.stdout)
+        logger.info(result.stdout)
 
     @pytest.mark.order(4)
     def test_list_jobs(self):
@@ -142,3 +169,96 @@ def test_cancel_job(self):
         result = self._execute_test_command(command)
         assert result.returncode == 0
         logger.info(result.stdout)
+    
+    @pytest.mark.order(7)
+    def test_cancel_job_with_quota(self):
+        command = [
+            "hyperpod",
+            "cancel-job",
+            "--job-name",
+            "hyperpod-cli-test-with-quota",
+        ]
+
+        result = self._execute_test_command(command)
+        assert result.returncode == 0
+        logger.info(result.stdout)
+
+    @pytest.mark.order(8)
+    def test_start_job_with_recipe(self):
+        command = [
+            "hyperpod",
+            "start-job",
+            "--recipe",
+            "fine-tuning/llama/hf_llama3_8b_seq8192_gpu",
+        ]
+
+        result = self._execute_test_command(command)
+        # wait for job to complete creation
+        time.sleep(240)
+        assert result.returncode == 0
+        logger.info(result.stdout)
+
+    @pytest.mark.order(9)
+    def test_start_job_with_recipe_and_override_parameters(self):
+        override_params = '''{
+            "recipes.run.name": "test-recipe-run",
+            "recipes.trainer.num_nodes": 1,
+            "instance_type": "g5.48xlarge"
+        }'''
+
+        command = [
+            "hyperpod",
+            "start-job",
+            "--recipe",
+            "fine-tuning/llama/hf_llama3_8b_seq8192_gpu",
+            "--override-parameters",
+            override_params,
+        ]
+
+        result = self._execute_test_command(command)
+        # wait for job to complete creation
+        time.sleep(240)
+        assert result.returncode == 0
+        logger.info(result.stdout)
+
+    @pytest.mark.order(10)
+    def test_get_job_with_recipe(self):
+        command = [
+            "hyperpod",
+            "get-job",
+            "--job-name",
+            "test-recipe-run",
+        ]
+
+        result = self._execute_test_command(command)
+        assert result.returncode == 0
+        logger.info(result.stdout)
+        assert "test-recipe-run" in str(result.stdout)
+        logger.info(result.stdout)
+
+    @pytest.mark.order(11)
+    def test_list_pods_with_recipe(self):
+        command = [
+            "hyperpod",
+            "list-pods",
+            "--job-name",
+            "test-recipe-run",
+        ]
+
+        result = self._execute_test_command(command)
+        assert result.returncode == 0
+        assert "test-recipe-run" in str(result.stdout)
+        logger.info(result.stdout)
+
+    @pytest.mark.order(12)
+    def test_cancel_job_with_recipe(self):
+        command = [
+            "hyperpod",
+            "cancel-job",
+            "--job-name",
+            "test-recipe-run",
+        ]
+
+        result = self._execute_test_command(command)
+        assert result.returncode == 0
+        logger.info(result.stdout)
diff --git a/test/unit_tests/clients/test_kubernetes_client.py b/test/unit_tests/clients/test_kubernetes_client.py
index 07618d3..f0fc88a 100644
--- a/test/unit_tests/clients/test_kubernetes_client.py
+++ b/test/unit_tests/clients/test_kubernetes_client.py
@@ -17,6 +17,7 @@
     Mock,
     mock_open,
     patch,
+    ANY,
 )
 
 import yaml
@@ -167,6 +168,32 @@ def test_set_context_success(
             "w",
         )
 
+        updated_kube_config_data = {
+            "contexts": [
+                {
+                    "name": "context1",
+                    "context": {
+                        "cluster": "cluster-1",
+                        "user": "user-1",
+                    },
+                },
+                {
+                    "name": "context2",
+                    "context": {
+                        "cluster": "cluster-2",
+                        "user": "user-2",
+                        "namespace": "default"
+                    },
+                },
+            ],
+            "current-context": "context2",
+        }
+
+        mock_safe_dump.assert_called_with(
+            updated_kube_config_data,
+            ANY,
+        )
+
     @patch("kubernetes.client.ApiClient")
     @patch("kubernetes.config.load_kube_config")
     @patch("yaml.safe_dump")
@@ -220,6 +247,59 @@ def test_set_context_success_other_namespace(
             "w",
         )
 
+    @patch("kubernetes.client.ApiClient")
+    @patch("kubernetes.config.load_kube_config")
+    @patch("yaml.safe_dump")
+    @patch("yaml.safe_load")
+    @patch(
+        "builtins.open",
+        new_callable=mock_open,
+        read_data=yaml.dump(KUBECONFIG_DATA),
+    )
+    def test_set_context_success_no_default(
+        self,
+        mock_open_file: Mock,
+        mock_safe_load: Mock,
+        mock_safe_dump: Mock,
+        mock_kube_config: Mock,
+        mock_client: Mock,
+    ):
+        mock_safe_load.return_value = KUBECONFIG_DATA
+        mock_safe_dump.return_value = None
+        mock_kube_config.return_value = None
+        mock_client.return_value = MagicMock()
+        test_client = KubernetesClient()
+        test_client.set_context("context2", None)
+        mock_open_file.assert_called_with(
+            os.path.expanduser(KUBE_CONFIG_DEFAULT_LOCATION),
+            "w",
+        )
+
+        updated_kube_config_data = {
+            "contexts": [
+                {
+                    "name": "context1",
+                    "context": {
+                        "cluster": "cluster-1",
+                        "user": "user-1",
+                    },
+                },
+                {
+                    "name": "context2",
+                    "context": {
+                        "cluster": "cluster-2",
+                        "user": "user-2",
+                    },
+                },
+            ],
+            "current-context": "context2",
+        }
+
+        mock_safe_dump.assert_called_with(
+            updated_kube_config_data,
+            ANY,
+        )
+
     @patch("kubernetes.client.ApiClient")
     @patch("kubernetes.config.load_kube_config")
     @patch("kubernetes.config.list_kube_config_contexts")
@@ -324,6 +404,20 @@ def test_get_apps_v1_api_error(
         with self.assertRaises(RuntimeError):
             test_client.get_apps_v1_api()
 
+    @patch("kubernetes.client.ApiClient")
+    @patch("kubernetes.config.load_kube_config")
+    def test_get_auth_v1_api(
+        self,
+        mock_kube_config: Mock,
+        mock_client: Mock,
+    ):
+        mock_kube_config.return_value = None
+        mock_client.return_value = MagicMock()
+        test_client = KubernetesClient()
+        test_client._kube_client = mock_client
+        auth_v1_api = test_client.get_auth_v1_api()
+        self.assertIsInstance(auth_v1_api, client.AuthorizationV1Api)
+
     @patch("kubernetes.config.load_kube_config")
     @patch(
         "kubernetes.client.CoreV1Api.list_node",
@@ -463,3 +557,98 @@ def test_list_pods_in_all_namespaces_with_labels_with_pagination(
         test_client = KubernetesClient()
         result = test_client.list_pods_in_all_namespaces_with_labels("kubeflow")
         self.assertEqual(2, len(result))
+
+    @patch("kubernetes.client.CoreV1Api.read_namespace")
+    def test_get_sagemaker_managed_namespace(
+        self,
+        mock_read_namespace
+    ):
+        test_client = KubernetesClient()
+        sm_managed_namespace = V1Namespace(
+            metadata=V1ObjectMeta(
+                name="test-sm-namespace", 
+                labels={
+                    "sagemaker.amazonaws.com/sagemaker-managed-queue": "true"
+                }
+            )
+        )
+        usr_managed_namespace = V1Namespace(
+            metadata=V1ObjectMeta(
+                name="test-ur-namespace", 
+            )
+        )
+        mock_read_namespace.side_effect = [sm_managed_namespace, usr_managed_namespace]
+
+        # verify SageMaker managed namespace would return
+        test_result = test_client.get_sagemaker_managed_namespace("test-sm-namespace")
+        self.assertEqual(test_result, sm_managed_namespace)
+
+        # verify user created namespace would return None
+        test_result = test_client.get_sagemaker_managed_namespace("test-ur-namespace")
+        self.assertIsNone(test_result)
+
+        # verify user created namespace would return None
+        test_result = test_client.get_sagemaker_managed_namespace(None)
+        self.assertIsNone(test_result)
+
+    @patch("kubernetes.client.CustomObjectsApi.patch_namespaced_custom_object")
+    def test_patch_workload(
+        self,
+        mock_patch_namespaced_custom_object,
+    ):
+        test_client = KubernetesClient()
+        test_client.patch_workload("test-workload", "test-namespace", "patch-body")
+
+        mock_patch_namespaced_custom_object.assert_called_with(
+            group="kueue.x-k8s.io",
+            version="v1beta1",
+            namespace="test-namespace",
+            plural="workloads",
+            name="test-workload",
+            body="patch-body"
+        )
+
+    @patch("kubernetes.client.CustomObjectsApi.list_namespaced_custom_object")
+    def test_get_workload_by_label(
+        self,
+        mock_list_namespaced_custom_object,
+    ):
+        test_client = KubernetesClient()
+        test_client.get_workload_by_label("label-selector", "test-namespace")
+
+        mock_list_namespaced_custom_object.assert_called_with(
+            group="kueue.x-k8s.io",
+            version="v1beta1",
+            namespace="test-namespace",
+            plural="workloads",
+            label_selector="label-selector",
+        )
+
+    @patch("kubernetes.client.CustomObjectsApi.list_cluster_custom_object")
+    def test_list_workload_priority_classes(
+        self,
+        mock_list_cluster_custom_object,
+    ):
+        test_client = KubernetesClient()
+        test_client.list_workload_priority_classes()
+
+        mock_list_cluster_custom_object.assert_called_with(
+            group="kueue.x-k8s.io",
+            version="v1beta1",
+            plural="workloadpriorityclasses",
+        )
+    
+    @patch("kubernetes.client.CustomObjectsApi.get_cluster_custom_object")
+    def test_get_cluster_queue(
+        self,
+        mock_get_cluster_custom_object,
+    ):
+        test_client = KubernetesClient()
+        test_client.get_cluster_queue("test-cluster-queue")
+
+        mock_get_cluster_custom_object.assert_called_with(
+            group="kueue.x-k8s.io",
+            version="v1beta1",
+            plural="clusterqueues",
+            name="test-cluster-queue",
+        )
diff --git a/test/unit_tests/service/test_cancel_training_job_service.py b/test/unit_tests/service/test_cancel_training_job_service.py
index 9191558..81596e0 100644
--- a/test/unit_tests/service/test_cancel_training_job_service.py
+++ b/test/unit_tests/service/test_cancel_training_job_service.py
@@ -23,7 +23,6 @@
 
 from kubernetes.client.rest import ApiException
 
-
 class CancelTrainingJobTest(unittest.TestCase):
     def setUp(self):
         self.mock_cancel_training_job = CancelTrainingJob()
@@ -73,3 +72,20 @@ def test_cancel_training_job_api_exception(
         with self.assertRaises(RuntimeError):
             self.mock_cancel_training_job.cancel_training_job("sample-job", None)
         mock_subprocess_run.assert_not_called()
+
+    @mock.patch("hyperpod_cli.service.discover_namespaces.DiscoverNamespaces.discover_accessible_namespace")
+    @mock.patch("subprocess.run")
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    def test_cancel_training_job_auto_discover_namespace(
+        self,
+        mock_kubernetes_client: mock.Mock,
+        mock_subprocess_run: mock.Mock,
+        mock_discover_accessible_namespace: mock.Mock,
+    ):
+        mock_kubernetes_client.return_value = self.mock_k8s_client
+        mock_discover_accessible_namespace.return_value = "discovered-namespace"
+        self.mock_k8s_client.get_current_context_namespace.return_value = None
+        self.mock_k8s_client.delete_training_job.return_value = {"status": "Success"}
+        result = self.mock_cancel_training_job.cancel_training_job("sample-job", None)
+        self.assertIsNone(result)
+        mock_subprocess_run.assert_called_once()
diff --git a/test/unit_tests/service/test_discover_namespaces.py b/test/unit_tests/service/test_discover_namespaces.py
new file mode 100644
index 0000000..8fb47df
--- /dev/null
+++ b/test/unit_tests/service/test_discover_namespaces.py
@@ -0,0 +1,194 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+#     http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+import unittest
+from unittest import mock
+from unittest.mock import MagicMock
+from kubernetes.client.rest import ApiException
+from hyperpod_cli.service.discover_namespaces import DiscoverNamespaces
+from concurrent.futures import Future
+from hyperpod_cli.clients.kubernetes_client import (
+    KubernetesClient,
+)
+
+class TestDiscoverNamespaces(unittest.TestCase):
+
+    def setUp(self):
+        self.mock_k8s_client = MagicMock(spec=KubernetesClient)
+
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.get_current_context_namespace")
+    def test_discover_accessible_sm_managed_namespace_explicit(
+        self, 
+        mock_get_current_context_namespace, 
+    ):
+        # Mock that the current context namespace is set
+        mock_get_current_context_namespace.return_value = "explicit-namespace"
+
+        # Instantiate the DiscoverNamespaces class
+        discover_ns = DiscoverNamespaces()
+
+        # Call discover_accessible_sm_managed_namespace with a resource attributes template
+        result = discover_ns.discover_accessible_namespace(resource_attributes_template={})
+
+        # Assert that the explicitly set namespace is returned
+        self.assertEqual(result, "explicit-namespace")
+
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    @mock.patch("hyperpod_cli.service.get_namespaces.GetNamespaces.get_sagemaker_managed_namespaces")
+    @mock.patch("hyperpod_cli.service.discover_namespaces.DiscoverNamespaces.get_namespaces_by_checking_access_permission")
+    def test_discover_accessible_sm_managed_namespace_success(
+        self, 
+        mock_get_namespaces_by_checking_access_permission, 
+        mock_get_sagemaker_managed_namespaces,
+        mock_kubernetes_client,
+    ):
+        # Mock clients
+        mock_kubernetes_client.return_value = self.mock_k8s_client
+        self.mock_k8s_client.get_current_context_namespace.return_value = None
+        
+        # Mock that no explicit namespace is set in the context
+        mock_get_sagemaker_managed_namespaces.return_value = ["namespace1"]
+        mock_get_namespaces_by_checking_access_permission.return_value = ["namespace1"]
+
+        # Instantiate the DiscoverNamespaces class
+        discover_ns = DiscoverNamespaces()
+
+        # Call discover_accessible_sm_managed_namespace with a resource attributes template
+        result = discover_ns.discover_accessible_namespace(resource_attributes_template={})
+
+        # Assert that the discovered namespace is returned
+        self.assertEqual(result, "namespace1")
+
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    @mock.patch("hyperpod_cli.service.get_namespaces.GetNamespaces.get_sagemaker_managed_namespaces")
+    @mock.patch("hyperpod_cli.service.discover_namespaces.DiscoverNamespaces.get_namespaces_by_checking_access_permission")
+    @mock.patch("sys.exit")
+    def test_discover_accessible_sm_managed_namespace_no_accessible(
+        self, 
+        mock_exit, 
+        mock_get_namespaces_by_checking_access_permission, 
+        mock_get_sagemaker_managed_namespaces,
+        mock_kubernetes_client,
+    ):
+        # Mock clients
+        mock_kubernetes_client.return_value = self.mock_k8s_client
+        self.mock_k8s_client.get_current_context_namespace.return_value = None
+        
+        # Mock that no namespaces are accessible
+        mock_get_sagemaker_managed_namespaces.return_value = ["namespace1"]
+        mock_get_namespaces_by_checking_access_permission.return_value = []
+        # Instead of exit, throw a SystemExit exception
+        mock_exit.side_effect = SystemExit
+
+        # Instantiate the DiscoverNamespaces class
+        discover_ns = DiscoverNamespaces()
+        
+        # Ensure SystemExit is raised
+        with self.assertRaises(SystemExit):
+            discover_ns.discover_accessible_namespace(resource_attributes_template={})
+
+        # Assert sys.exit is called with status code 1
+        mock_exit.assert_called_once_with(1)
+
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    @mock.patch("hyperpod_cli.service.get_namespaces.GetNamespaces.get_sagemaker_managed_namespaces")
+    @mock.patch("hyperpod_cli.service.discover_namespaces.DiscoverNamespaces.get_namespaces_by_checking_access_permission")
+    @mock.patch("sys.exit")
+    def test_discover_accessible_sm_managed_namespace_multiple_accessible(
+        self, 
+        mock_exit, 
+        mock_get_namespaces_by_checking_access_permission, 
+        mock_get_sagemaker_managed_namespaces,
+        mock_kubernetes_client,
+    ):
+        # Mock clients
+        mock_kubernetes_client.return_value = self.mock_k8s_client
+        self.mock_k8s_client.get_current_context_namespace.return_value = None
+
+        # Mock that multiple namespaces are accessible
+        mock_get_sagemaker_managed_namespaces.return_value = ["namespace1", "namespace2"]
+        mock_get_namespaces_by_checking_access_permission.return_value = ["namespace1", "namespace2"]
+        # Instead of exit, throw a SystemExit exception
+        mock_exit.side_effect = SystemExit
+
+        # Instantiate the DiscoverNamespaces class
+        discover_ns = DiscoverNamespaces()
+
+        # Ensure SystemExit is raised
+        with self.assertRaises(SystemExit):
+            discover_ns.discover_accessible_namespace(resource_attributes_template={})
+
+        # Assert sys.exit is called with status code 1 for multiple namespaces
+        mock_exit.assert_called_once_with(1)
+
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    @mock.patch("hyperpod_cli.service.get_namespaces.GetNamespaces.get_sagemaker_managed_namespaces")
+    @mock.patch("sys.exit")
+    def test_discover_accessible_sm_managed_namespace_api_exception(
+        self, 
+        mock_exit, 
+        mock_get_sagemaker_managed_namespaces,
+        mock_kubernetes_client,
+    ):
+        # Mock clients
+        mock_kubernetes_client.return_value = self.mock_k8s_client
+        self.mock_k8s_client.get_current_context_namespace.return_value = None
+
+        # Simulate a 403 Forbidden ApiException
+        mock_get_sagemaker_managed_namespaces.side_effect = ApiException(status=403)
+        # Instead of exit, throw a SystemExit exception
+        mock_exit.side_effect = SystemExit
+
+        # Instantiate the DiscoverNamespaces class
+        discover_ns = DiscoverNamespaces()
+
+         # Ensure SystemExit is raised
+        with self.assertRaises(SystemExit):
+            discover_ns.discover_accessible_namespace(resource_attributes_template={})
+
+        # Assert sys.exit is called with status code 1
+        mock_exit.assert_called_once_with(1)
+
+    @mock.patch("hyperpod_cli.service.self_subject_access_review.SelfSubjectAccessReview.self_subject_access_review")
+    def test_get_namespaces_by_checking_access_permission(
+        self, 
+        mock_self_subject_access_review,
+    ):
+        # Mock the SelfSubjectAccessReview response
+        mock_future = Future()
+        mock_response = MagicMock()
+        mock_response.status.allowed = True
+        mock_response.spec.resource_attributes = MagicMock(namespace="namespace1")
+        mock_future.set_result(mock_response)
+
+        mock_self_subject_access_review.return_value = mock_future.result()
+
+        # Mock the resource attributes object
+        resource_attributes_template = MagicMock()
+        resource_attributes_template.namespace = None
+
+        # Instantiate the DiscoverNamespaces class
+        discover_ns = DiscoverNamespaces()
+
+        # Call get_namespaces_by_checking_access_permission
+        namespaces = discover_ns.get_namespaces_by_checking_access_permission(
+            namespaces=["namespace1"],
+            resource_attributes_template=resource_attributes_template,
+            max_workers=1
+        )
+
+        # Assert that the namespace is correctly identified as accessible
+        self.assertEqual(namespaces, ["namespace1"])
+
+
+if __name__ == "__main__":
+    unittest.main()
diff --git a/test/unit_tests/service/test_get_logs_service.py b/test/unit_tests/service/test_get_logs_service.py
index 5bcda27..e982896 100644
--- a/test/unit_tests/service/test_get_logs_service.py
+++ b/test/unit_tests/service/test_get_logs_service.py
@@ -103,3 +103,25 @@ def test_get_logs_without_namespace_api_exception(
         )
         with self.assertRaises(RuntimeError):
             self.mock_get_logs.get_training_job_logs("sample-job", "test-pod", None)
+
+    @mock.patch("hyperpod_cli.service.discover_namespaces.DiscoverNamespaces.discover_accessible_namespace")
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    @mock.patch("hyperpod_cli.service.list_pods.ListPods")
+    @mock.patch("hyperpod_cli.service.list_pods.ListPods.list_pods_for_training_job")
+    def test_get_logs_auto_discover_namespace(
+        self,
+        mock_list_training_job_pods_service_with_list_pods: mock.Mock,
+        mock_list_training_job_pods_service: mock.Mock,
+        mock_kubernetes_client: mock.Mock,
+        mock_discover_accessible_namespace: mock.Mock,
+    ):
+        mock_kubernetes_client.return_value = self.mock_k8s_client
+        mock_list_training_job_pods_service.return_value = self.mock_list_pods_service
+        self.mock_k8s_client.get_current_context_namespace.return_value = None
+        mock_discover_accessible_namespace.return_value = "discovered-namespace"
+        mock_list_training_job_pods_service_with_list_pods.return_value = ["test-pod"]
+        self.mock_k8s_client.get_logs_for_pod.return_value = "test logs"
+        result = self.mock_get_logs.get_training_job_logs(
+            "sample-job", "test-pod", None
+        )
+        self.assertIn("test logs", result)
diff --git a/test/unit_tests/service/test_get_namespaces.py b/test/unit_tests/service/test_get_namespaces.py
new file mode 100644
index 0000000..7317968
--- /dev/null
+++ b/test/unit_tests/service/test_get_namespaces.py
@@ -0,0 +1,102 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+#     http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+import unittest
+from unittest import mock
+from unittest.mock import MagicMock
+
+from hyperpod_cli.clients.kubernetes_client import KubernetesClient
+from hyperpod_cli.service.get_namespaces import GetNamespaces
+
+
+class TestGetNamespaces(unittest.TestCase):
+    
+    def setUp(self):
+        self.mock_k8s_client = MagicMock(spec=KubernetesClient)
+        self.mock_get_namespaces = GetNamespaces()
+
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    def test_get_namespaces_success(self, mock_kubernetes_client):
+        mock_core_v1_api = MagicMock()
+        mock_kubernetes_client.return_value = self.mock_k8s_client
+        self.mock_k8s_client.get_core_v1_api.return_value = mock_core_v1_api
+
+        mock_namespace_list = MagicMock()
+        mock_namespace1 = MagicMock()
+        mock_namespace1.metadata.name = "namespace1"
+        
+        mock_namespace2 = MagicMock()
+        mock_namespace2.metadata.name = "namespace2"
+        mock_namespace_list.items = [mock_namespace1, mock_namespace2]
+
+        mock_namespace_list.metadata._continue = None
+        mock_core_v1_api.list_namespace.return_value = mock_namespace_list
+
+        namespaces = self.mock_get_namespaces.get_namespaces()
+
+        self.assertEqual(namespaces, ["namespace1", "namespace2"])
+
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    def test_get_namespaces_pagination(self, mock_kubernetes_client):
+        mock_kubernetes_client.return_value = self.mock_k8s_client
+        mock_core_v1_api = MagicMock()
+        self.mock_k8s_client.get_core_v1_api.return_value = mock_core_v1_api
+
+        # First response with a continue token
+        first_response = MagicMock()
+        mock_namespace1 = MagicMock()
+        mock_namespace1.metadata.name = "namespace1"
+        first_response.items = [mock_namespace1]
+        first_response.metadata._continue = "next-token"
+
+        # Second response without a continue token
+        second_response = MagicMock()
+        mock_namespace2 = MagicMock()
+        mock_namespace2.metadata.name = "namespace2"
+        second_response.items = [mock_namespace2]
+        second_response.metadata._continue = None
+
+        mock_core_v1_api.list_namespace.side_effect = [
+            first_response, second_response
+        ]
+
+        namespaces = self.mock_get_namespaces.get_namespaces()
+        self.assertEqual(namespaces, ["namespace1", "namespace2"])
+        self.assertEqual(mock_core_v1_api.list_namespace.call_count, 2)
+
+
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    def test_get_sagemaker_managed_namespaces(self, mock_kubernetes_client):
+        mock_kubernetes_client.return_value = self.mock_k8s_client
+        mock_core_v1_api = MagicMock()
+        self.mock_k8s_client.get_core_v1_api.return_value = mock_core_v1_api
+
+        first_response = MagicMock()
+        mock_namespace1 = MagicMock()
+        mock_namespace1.metadata.name = "namespace1"
+        first_response.items = [mock_namespace1]
+        first_response.metadata._continue = None
+
+        mock_core_v1_api.list_namespace.side_effect = [
+            first_response
+        ]
+
+        namespaces = self.mock_get_namespaces.get_sagemaker_managed_namespaces()
+        self.assertEqual(namespaces, ["namespace1"])
+        mock_core_v1_api.list_namespace.assert_called_once_with(
+            limit=200, 
+            label_selector="sagemaker.amazonaws.com/sagemaker-managed-queue=true"
+        )
+
+
+if __name__ == "__main__":
+    unittest.main()
diff --git a/test/unit_tests/service/test_get_training_job_service.py b/test/unit_tests/service/test_get_training_job_service.py
index dc08fb8..e48d0b9 100644
--- a/test/unit_tests/service/test_get_training_job_service.py
+++ b/test/unit_tests/service/test_get_training_job_service.py
@@ -17,6 +17,8 @@
 from hyperpod_cli.clients.kubernetes_client import (
     KubernetesClient,
 )
+from kubernetes.client import V1ResourceAttributes
+
 from hyperpod_cli.service.get_training_job import (
     GetTrainingJob,
 )
@@ -65,6 +67,38 @@ def test_get_training_job_without_namespace(
         result = self.mock_get_training_job.get_training_job("sample-job", None, None)
         self.assertIn("pytorch-simple", result)
 
+    @mock.patch("hyperpod_cli.service.discover_namespaces.DiscoverNamespaces.discover_accessible_namespace")
+    @mock.patch("hyperpod_cli.utils.get_cluster_console_url")
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    def test_get_training_job_namespace_auto_discover(
+        self,
+        mock_kubernetes_client: mock.Mock,
+        mock_current_hyperpod_context: mock.Mock,
+        mock_discover_accessible_namespace: mock.Mock,
+    ):
+        mock_kubernetes_client.return_value = self.mock_k8s_client
+        mock_current_hyperpod_context.return_value = SAMPLE_HYPERPOD_CLUSTER_URL
+        mock_discover_accessible_namespace.return_value = "discovered-namespace"
+
+        self.mock_k8s_client.get_current_context_namespace.return_value = None
+        self.mock_k8s_client.get_job.return_value = {
+            "metadata": {"name": "pytorch-simple"}
+        }
+        result = self.mock_get_training_job.get_training_job("sample-job", None, None)
+        # Ensure that we are using correct resource attributes for auto-discover
+        mock_discover_accessible_namespace.assert_called_once_with(
+            V1ResourceAttributes(
+                verb="get",
+                group="kubeflow.org",
+                resource="pytorchjobs",
+            )
+        )
+        self.mock_k8s_client.get_job.assert_called_once_with(
+            job_name="sample-job",
+            namespace="discovered-namespace",
+        )
+        self.assertIn("pytorch-simple", result)
+
     @mock.patch("hyperpod_cli.utils.get_cluster_console_url")
     @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
     def test_get_training_job_with_verbose(
@@ -141,7 +175,6 @@ def test_get_training_job_with_verbose_with_values(
             "status": "test_status",
         }
         result = self.mock_get_training_job.get_training_job("sample-job", None, True)
-        print(result)
         self.assertIn("pytorch-simple", result)
 
     @mock.patch("hyperpod_cli.utils.get_cluster_console_url")
@@ -161,7 +194,6 @@ def test_get_training_job_with_verbose_without_metadata(
             "status": "test_status",
         }
         result = self.mock_get_training_job.get_training_job("sample-job", None, True)
-        print(result)
         self.assertIn("test_status", result)
 
     @mock.patch("hyperpod_cli.utils.get_cluster_console_url")
diff --git a/test/unit_tests/service/test_list_pods_service.py b/test/unit_tests/service/test_list_pods_service.py
index 37e23d9..6159ea8 100644
--- a/test/unit_tests/service/test_list_pods_service.py
+++ b/test/unit_tests/service/test_list_pods_service.py
@@ -217,3 +217,17 @@ def test_list_pods_and_get_requested_resources_should_skip_if_node_name_is_empty
         ]
         result = self.mock_list_pods.list_pods_and_get_requested_resources_group_by_node_name()
         self.assertEqual(len(result), 0)
+    
+    @mock.patch("hyperpod_cli.service.discover_namespaces.DiscoverNamespaces.discover_accessible_namespace")
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    def test_list_pods_auto_discover_namespace(
+        self,
+        mock_kubernetes_client: mock.Mock,
+        mock_discover_accessible_namespace: mock.Mock,
+    ):
+        mock_kubernetes_client.return_value = self.mock_k8s_client
+        self.mock_k8s_client.get_current_context_namespace.return_value = None
+        mock_discover_accessible_namespace.return_value = "discovered-namespace"
+        self.mock_k8s_client.list_pods_with_labels.return_value = SAMPLE_OUTPUT
+        result = self.mock_list_pods.list_pods_for_training_job("test-job", None, False)
+        self.assertEqual(2, len(result))
diff --git a/test/unit_tests/service/test_list_training_jobs_service.py b/test/unit_tests/service/test_list_training_jobs_service.py
index 61fccc0..b0d7f37 100644
--- a/test/unit_tests/service/test_list_training_jobs_service.py
+++ b/test/unit_tests/service/test_list_training_jobs_service.py
@@ -17,6 +17,7 @@
 from hyperpod_cli.clients.kubernetes_client import (
     KubernetesClient,
 )
+from kubernetes.client import V1ResourceAttributes
 from hyperpod_cli.service.list_training_jobs import (
     ListTrainingJobs,
 )
@@ -39,6 +40,19 @@
                     }
                 ],
             },
+            "spec": {
+                "pytorchReplicaSpecs": {
+                    "Worker": {
+                        "template": {
+                            "metadata": {
+                                "labels": {
+                                    "kueue.x-k8s.io/priority-class": "priority-1"
+                                }
+                            }
+                        }
+                    }
+                }
+            }
         },
         {
             "metadata": {
@@ -48,16 +62,27 @@
             "status": {
                 "startTime": "test-time1",
                 "conditions": [
-                    {
-                        "type": "Running",
-                        "lastTransitionTime": "2024-08-27T22:47:57Z",
-                    },
                     {
                         "type": "Created",
                         "lastTransitionTime": "2023-08-27T22:47:57Z",
                     },
+                    {
+                        "type": "Running",
+                        "lastTransitionTime": "2024-08-27T22:47:57Z",
+                    },
                 ],
             },
+            "spec": {
+                "pytorchReplicaSpecs": {
+                    "Worker": {
+                        "template": {
+                            "spec": {
+                                "priorityClassName": "priority-2"
+                            }
+                        }
+                    }
+                }
+            }
         },
         {
             "metadata": {
@@ -73,6 +98,22 @@
                     }
                 ],
             },
+            "spec": {
+                "pytorchReplicaSpecs": {
+                    "Worker": {
+                        "template": {
+                            "spec": {
+                                "priorityClassName": "priority-3",
+                            },
+                            "metadata": {
+                                "labels": {
+                                    "kueue.x-k8s.io/priority-class": "priority-4"
+                                }
+                            }
+                        }
+                    }
+                }
+            }
         },
     ]
 }
@@ -115,6 +156,7 @@ def test_list_training_jobs_with_namespace(
         )
         self.assertIn("test-name", result)
         self.assertIn("test-name1", result)
+        self.assertIn("Running", result)
 
     @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
     def test_list_training_jobs_without_namespace(
@@ -127,6 +169,37 @@ def test_list_training_jobs_without_namespace(
         result = self.mock_list_training_jobs.list_training_jobs(None, None, None)
         self.assertIn("test-name", result)
         self.assertIn("test-name1", result)
+        self.assertIn("Running", result)
+
+    @mock.patch("hyperpod_cli.service.discover_namespaces.DiscoverNamespaces.discover_accessible_namespace")
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    def test_list_training_jobs_with_namespace_auto_discover(
+        self,
+        mock_kubernetes_client: mock.Mock,
+        mock_discover_accessible_namespace: mock.Mock,
+    ):
+        mock_kubernetes_client.return_value = self.mock_k8s_client
+        mock_discover_accessible_namespace.return_value = "discovered-namespace"
+        self.mock_k8s_client.get_current_context_namespace.return_value = None
+        self.mock_k8s_client.list_training_jobs.return_value = SAMPLE_OUTPUT
+        result = self.mock_list_training_jobs.list_training_jobs(None, None, None)
+        mock_discover_accessible_namespace.assert_called_once_with(
+            V1ResourceAttributes(
+                verb="list",
+                group="kubeflow.org",
+                resource="pytorchjobs",
+            )
+        )
+        self.mock_k8s_client.list_training_jobs.assert_called_once_with(
+            namespace="discovered-namespace",
+            label_selector=None,
+        )
+
+        self.assertIn("test-name", result)
+        self.assertIn("test-name1", result)
+        self.assertIn("priority-1", result)
+        self.assertIn("priority-2", result)
+        self.assertIn("priority-4", result)
 
     @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
     def test_list_training_jobs_without_namespace_no_jobs(
diff --git a/test/unit_tests/service/test_self_subject_access_review.py b/test/unit_tests/service/test_self_subject_access_review.py
new file mode 100644
index 0000000..cd142b3
--- /dev/null
+++ b/test/unit_tests/service/test_self_subject_access_review.py
@@ -0,0 +1,60 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+#     http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+import unittest
+from unittest import mock
+from unittest.mock import MagicMock
+from kubernetes.client import V1SelfSubjectAccessReview
+from kubernetes.client import AuthorizationV1Api
+from hyperpod_cli.service.self_subject_access_review import SelfSubjectAccessReview
+
+
+class TestSelfSubjectAccessReview(unittest.TestCase):
+
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    def test_self_subject_access_review_success(self, mock_kubernetes_client):
+        # Mock the Kubernetes API response
+        mock_auth_v1_api = MagicMock(spec=AuthorizationV1Api)
+        mock_kubernetes_client().get_auth_v1_api.return_value = mock_auth_v1_api
+        
+        # Create a mock response for create_self_subject_access_review
+        mock_response = MagicMock(spec=V1SelfSubjectAccessReview)
+        mock_response.status.allowed = True
+        mock_auth_v1_api.create_self_subject_access_review.return_value = mock_response
+
+        # Set up resource attributes for the access review
+        resource_attributes = {
+            "namespace": "test-namespace",
+            "verb": "create",
+            "resource": "pods"
+        }
+
+        # Instantiate the SelfSubjectAccessReview service
+        service = SelfSubjectAccessReview()
+
+        # Call the self_subject_access_review method
+        response = service.self_subject_access_review(resource_attributes=resource_attributes)
+
+        # Check that the response indicates the action is allowed
+        self.assertTrue(response.status.allowed)
+
+        # Assert that the Kubernetes API was called correctly
+        mock_auth_v1_api.create_self_subject_access_review.assert_called_once()
+
+        # Capture and inspect the arguments passed in the call
+        args, kwargs = mock_auth_v1_api.create_self_subject_access_review.call_args
+        self.assertIsInstance(kwargs['body'], V1SelfSubjectAccessReview)
+        self.assertEqual(kwargs['body'].spec.resource_attributes['namespace'], 'test-namespace')
+
+
+if __name__ == "__main__":
+    unittest.main()
diff --git a/test/unit_tests/test_cluster.py b/test/unit_tests/test_cluster.py
index 32f9d8a..8fdd605 100644
--- a/test/unit_tests/test_cluster.py
+++ b/test/unit_tests/test_cluster.py
@@ -19,7 +19,10 @@
 
 from click.testing import CliRunner
 from kubernetes import client
-
+from kubernetes.client import (
+    V1Namespace, 
+    V1ObjectMeta,
+)
 from hyperpod_cli.clients.kubernetes_client import (
     KubernetesClient,
 )
@@ -939,6 +942,153 @@ def test_list_clusters_skipped_not_eks_clusters(
         self.assertNotIn("cluster-1", result.output)
         self.assertNotIn("cluster-2", result.output)
 
+    @mock.patch("kubernetes.config.load_kube_config")
+    @mock.patch("boto3.Session")
+    @mock.patch(
+        "hyperpod_cli.commands.cluster.ClusterValidator.validate_aws_credential"
+    )
+    @mock.patch(
+        "hyperpod_cli.commands.cluster.ClusterValidator.validate_cluster_and_get_eks_arn"
+    )
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    @mock.patch("subprocess.run")
+    def test_get_clusters_with_sm_managed_namespace(
+        self,
+        mock_subprocess_run: mock.Mock,
+        mock_kubernetes_client: mock.Mock,
+        mock_validate_cluster_and_get_eks_arn: mock.Mock,
+        mock_validate_aws_credentials: mock.Mock,
+        mock_session: mock.Mock,
+        mock_load_kube_config: mock.Mock,
+    ):
+        self.mock_k8s_client.list_node_with_temp_config.return_value = (
+            _generate_nodes_list()
+        )
+        mock_kubernetes_client.return_value = self.mock_k8s_client
+
+        mock_validate_aws_credentials.validate_aws_credential.return_value = None
+        mock_validate_cluster_and_get_eks_arn.return_value = (
+            "arn:aws:eks:us-west-2:123456789012:cluster/cluster-1"
+        )
+
+        mock_load_kube_config.return_value = None
+        mock_subprocess_run.return_value = None
+        self.mock_k8s_client.get_sagemaker_managed_namespace.return_value = V1Namespace(
+            metadata=V1ObjectMeta(
+                name="test-namespace",
+                labels={
+                    "sagemaker.amazonaws.com/sagemaker-managed-queue": "true",
+                    "sagemaker.amazonaws.com/quota-allocation-id": "test-team",
+                },
+            ),
+        )
+        self.mock_k8s_client.get_cluster_queue.return_value = _generate_get_cluster_queue_response()
+
+        self.mock_sm_client.describe_cluster.return_value = {
+            "Orchestrator": {
+                "Eks": {
+                    "ClusterArn": "arn:aws:eks:us-west-2:123456789012:cluster/cluster-1"
+                }
+            }
+        }
+        self.mock_sm_client.list_clusters.return_value = (
+            {
+                "ClusterSummaries": [
+                    {"ClusterName": "cluster-1"},
+                ]
+            }
+        )
+        self.mock_session.client.return_value = self.mock_sm_client
+        mock_session.return_value = self.mock_session
+
+        result = self.runner.invoke(
+            get_clusters,
+            [
+                "-n",
+                "test-namespace",
+            ],
+            catch_exceptions=False,
+        )
+        self.assertEqual(result.exit_code, 0)
+        self.assertIn("cluster-1", result.output)
+        # For instance-type-1
+        self.assertIn("test-namespace", result.output)
+        self.assertIn('"TotalAcceleratorDevices": 1', result.output)
+        self.assertIn('"AvailableAcceleratorDevices": 0', result.output)
+        # For instance-type-2
+        self.assertIn('"TotalAcceleratorDevices": "N/A"', result.output)
+        self.assertIn('"AvailableAcceleratorDevices": "N/A"', result.output)
+        # Expect JSON output
+        json.loads(result.output)
+
+    @mock.patch("kubernetes.config.load_kube_config")
+    @mock.patch("boto3.Session")
+    @mock.patch(
+        "hyperpod_cli.commands.cluster.ClusterValidator.validate_aws_credential"
+    )
+    @mock.patch(
+        "hyperpod_cli.commands.cluster.ClusterValidator.validate_cluster_and_get_eks_arn"
+    )
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    @mock.patch("subprocess.run")
+    def test_get_clusters_with_not_sm_managed_namespace(
+        self,
+        mock_subprocess_run: mock.Mock,
+        mock_kubernetes_client: mock.Mock,
+        mock_validate_cluster_and_get_eks_arn: mock.Mock,
+        mock_validate_aws_credentials: mock.Mock,
+        mock_session: mock.Mock,
+        mock_load_kube_config: mock.Mock,
+    ):
+        self.mock_k8s_client.list_node_with_temp_config.return_value = (
+            _generate_nodes_list()
+        )
+        mock_kubernetes_client.return_value = self.mock_k8s_client
+
+        mock_validate_aws_credentials.validate_aws_credential.return_value = None
+        mock_validate_cluster_and_get_eks_arn.return_value = (
+            "arn:aws:eks:us-west-2:123456789012:cluster/cluster-1"
+        )
+
+        mock_load_kube_config.return_value = None
+        mock_subprocess_run.return_value = None
+        self.mock_k8s_client.get_sagemaker_managed_namespace.return_value = None
+
+        self.mock_sm_client.describe_cluster.return_value = {
+            "Orchestrator": {
+                "Eks": {
+                    "ClusterArn": "arn:aws:eks:us-west-2:123456789012:cluster/cluster-1"
+                }
+            }
+        }
+        self.mock_sm_client.list_clusters.return_value = (
+            {
+                "ClusterSummaries": [
+                    {"ClusterName": "cluster-1"},
+                ]
+            }
+        )
+        self.mock_session.client.return_value = self.mock_sm_client
+        mock_session.return_value = self.mock_session
+
+        result = self.runner.invoke(
+            get_clusters,
+            [
+                "-n",
+                "test-namespace",
+            ],
+            catch_exceptions=False,
+        )
+        self.assertEqual(result.exit_code, 0)
+        self.assertIn("cluster-1", result.output)
+
+        # For both instance-type-1 and instance-type-2
+        self.assertIn("test-namespace", result.output)
+        self.assertIn('"TotalAcceleratorDevices": "N/A"', result.output)
+        self.assertIn('"AvailableAcceleratorDevices": "N/A"', result.output)
+        # Expect JSON output
+        json.loads(result.output)
+
 
 def _generate_nodes_list():
     return [
@@ -1045,6 +1195,42 @@ def _generate_get_clusters_response():
         ]
     }
 
+def _generate_get_cluster_queue_response():
+    return {
+        "kind": "ClusterQueue",
+        "metadata": {
+            "name": "test-queue",
+        },
+        "spec": {
+            "resourceGroups": [
+                {
+                    "coveredResources": ["cpu", "memory", "nvidia.com/gpu"],
+                    "flavors": [
+                        {
+                            "name": "instance-type-1",
+                            "resources": [
+                                {"name": "cpu", "nominalQuota": 10},
+                                {"name": "memory", "nominalQuota": "10Gi"},
+                                {"name": "nvidia.com/gpu", "nominalQuota": 1},
+                            ]
+                        },
+                    ]
+                }
+            ],
+        },
+        "status": {
+            "flavorsUsage": [
+                {
+                    "name": "instance-type-1",
+                    "resources": [
+                        {"name": "cpu", "total": 5, "borrowed": 1},
+                        {"name": "memory", "total": "4Gi", "borrowed": "0Gi"},
+                        {"name": "nvidia.com/gpu", "total": 1, "borrowed": 0}
+                    ]
+                }
+            ]
+        }
+    }
 
 def _generate_get_clusters_response_over_maximum():
     cluster_list = [{"ClusterName": f"cluster-{i+1}"} for i in range(100)]
diff --git a/test/unit_tests/test_job.py b/test/unit_tests/test_job.py
index a5770c6..098aff8 100644
--- a/test/unit_tests/test_job.py
+++ b/test/unit_tests/test_job.py
@@ -13,17 +13,23 @@
 import unittest
 import click
 import subprocess
+import pytest
 import os
+import yaml
 from unittest import mock
 from unittest.mock import MagicMock, mock_open
 
 from click.testing import CliRunner
-
+from kubernetes.client import (
+    V1Namespace, 
+    V1ObjectMeta,
+)
 from hyperpod_cli.commands.job import (
     cancel_job,
     get_job,
     list_jobs,
     list_pods,
+    patch_job,
     start_job,
     suppress_standard_output_context,
     validate_only_config_file_argument,
@@ -52,6 +58,9 @@ def setUp(self):
         self.mock_list_jobs = MagicMock(spec=ListTrainingJobs)
         self.list_pods = MagicMock(spec=ListPods)
 
+        # Patch gettext for now because see some issues locally for localization
+        mock.patch("gettext.dgettext", lambda domain, message: message).start()
+        
     @mock.patch("hyperpod_cli.service.get_training_job.GetTrainingJob")
     @mock.patch("hyperpod_cli.service.get_training_job.GetTrainingJob.get_training_job")
     def test_get_job_happy_case(
@@ -124,15 +133,6 @@ def test_get_job_happy_case_with_namespace_and_verbose(
         self.assertEqual(result.exit_code, 0)
         self.assertIn("example-job", result.output)
 
-    def test_get_job_error_missing_name_option(
-        self,
-    ):
-        result = self.runner.invoke(get_job, ["example-job"])
-        self.assertIn(
-            "Missing option '--job-name'",
-            result.output,
-        )
-
     @mock.patch("hyperpod_cli.service.get_training_job.GetTrainingJob")
     @mock.patch("hyperpod_cli.service.get_training_job.GetTrainingJob.get_training_job")
     def test_get_job_when_subprocess_command_gives_exception(
@@ -441,13 +441,7 @@ def test_cancel_job_when_subprocess_command_gives_exception(
             result.output,
         )
 
-    @mock.patch("hyperpod_cli.commands.job.initialize_config_dir")
-    @mock.patch("hyperpod_cli.commands.job.compose")
-    @mock.patch("hyperpod_cli.commands.job.customer_launcher")
-    @mock.patch(
-        "builtins.open",
-        new_callable=mock.mock_open,
-    )
+    @mock.patch('subprocess.run')
     @mock.patch("yaml.dump")
     @mock.patch("os.path.exists", return_value=True)
     @mock.patch("os.remove", return_value=None)
@@ -459,24 +453,24 @@ def test_start_job_with_cli_args(
         self,
         mock_boto3,
         mock_validator_cls,
-        mock_kubernetes_client: mock.Mock,
+        mock_kubernetes_client,
         mock_get_console_link,
         mock_remove,
         mock_exists,
         mock_yaml_dump,
-        mock_file,
-        mock_main,
-        mock_compose,
-        mock_initialize_config_dir,
+        mock_subprocess_run,
     ):
         mock_validator = mock_validator_cls.return_value
         mock_validator.validate_aws_credential.return_value = True
         mock_kubernetes_client.get_current_context_namespace.return_value = "kubeflow"
         mock_get_console_link.return_value = "test-console-link"
         mock_yaml_dump.return_value = None
-        mock_main.return_value = None
-        mock_compose.return_value = None
-        mock_initialize_config_dir.return_value.__enter__.return_value = None
+        mock_subprocess_run.return_value = subprocess.CompletedProcess(
+            args=['some_command'],  # Simulate the command arguments
+            returncode=0,           # Simulate a successful command
+            stdout='Command executed successfully',  # Simulate standard output
+            stderr=''               # Simulate no errors
+        )
         result = self.runner.invoke(
             start_job,
             [
@@ -491,42 +485,44 @@ def test_start_job_with_cli_args(
                 "--entry-script",
                 "/opt/train/src/train.py",
             ],
+            catch_exceptions=False
         )
+        print(f"Exit code: {result.exit_code}")
+        print(f"Output: {result.output}")
+        if result.exception:
+            print(f"Exception: {result.exception}")
         self.assertEqual(result.exit_code, 0)
 
-    @mock.patch("hyperpod_cli.commands.job.initialize_config_dir")
-    @mock.patch("hyperpod_cli.commands.job.compose")
-    @mock.patch("hyperpod_cli.commands.job.customer_launcher")
-    @mock.patch(
-        "builtins.open",
-        new_callable=mock.mock_open,
-    )
+    @mock.patch('subprocess.run')
     @mock.patch("yaml.dump")
     @mock.patch("os.path.exists", return_value=True)
     @mock.patch("os.remove", return_value=None)
     @mock.patch("hyperpod_cli.utils.get_cluster_console_url")
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
     @mock.patch("hyperpod_cli.commands.job.JobValidator")
     @mock.patch("boto3.Session")
     def test_start_job_with_cli_args_with_namespace(
         self,
         mock_boto3,
         mock_validator_cls,
+        mock_kubernetes_client,
         mock_get_console_link,
         mock_remove,
         mock_exists,
         mock_yaml_dump,
-        mock_file,
-        mock_main,
-        mock_compose,
-        mock_initialize_config_dir,
+        mock_subprocess_run,
     ):
         mock_validator = mock_validator_cls.return_value
         mock_validator.validate_aws_credential.return_value = True
+        mock_kubernetes_client.get_current_context_namespace.return_value = "kubeflow"
         mock_get_console_link.return_value = "test-console-link"
         mock_yaml_dump.return_value = None
-        mock_main.return_value = None
-        mock_compose.return_value = None
-        mock_initialize_config_dir.return_value.__enter__.return_value = None
+        mock_subprocess_run.return_value = subprocess.CompletedProcess(
+            args=['some_command'],  # Simulate the command arguments
+            returncode=0,           # Simulate a successful command
+            stdout='Command executed successfully',  # Simulate standard output
+            stderr=''               # Simulate no errors
+        )
         result = self.runner.invoke(
             start_job,
             [
@@ -546,13 +542,7 @@ def test_start_job_with_cli_args_with_namespace(
         )
         self.assertEqual(result.exit_code, 0)
 
-    @mock.patch("hyperpod_cli.commands.job.initialize_config_dir")
-    @mock.patch("hyperpod_cli.commands.job.compose")
-    @mock.patch("hyperpod_cli.commands.job.customer_launcher")
-    @mock.patch(
-        "builtins.open",
-        new_callable=mock.mock_open,
-    )
+    @mock.patch('subprocess.run')
     @mock.patch("yaml.dump")
     @mock.patch("os.path.exists", return_value=True)
     @mock.patch("os.remove", return_value=None)
@@ -571,19 +561,19 @@ def test_start_job_with_cli_args_debug_mode(
         mock_remove,
         mock_exists,
         mock_yaml_dump,
-        mock_file,
-        mock_main,
-        mock_compose,
-        mock_initialize_config_dir,
+        mock_subprocess_run,
     ):
         mock_validator = mock_validator_cls.return_value
         mock_validator.validate_aws_credential.return_value = True
         mock_kubernetes_client.get_current_context_namespace.return_value = "kubeflow"
         mock_get_console_link.return_value = "test-console-link"
         mock_yaml_dump.return_value = None
-        mock_main.return_value = None
-        mock_compose.return_value = None
-        mock_initialize_config_dir.return_value.__enter__.return_value = None
+        mock_subprocess_run.return_value = subprocess.CompletedProcess(
+            args=['some_command'],  # Simulate the command arguments
+            returncode=0,           # Simulate a successful command
+            stdout='Command executed successfully',  # Simulate standard output
+            stderr=''               # Simulate no errors
+        )
         result = self.runner.invoke(
             start_job,
             [
@@ -603,13 +593,7 @@ def test_start_job_with_cli_args_debug_mode(
         self.assertEqual(result.exit_code, 0)
         mock_debug.assert_called()
 
-    @mock.patch("hyperpod_cli.commands.job.initialize_config_dir")
-    @mock.patch("hyperpod_cli.commands.job.compose")
-    @mock.patch("hyperpod_cli.commands.job.customer_launcher")
-    @mock.patch(
-        "builtins.open",
-        new_callable=mock.mock_open,
-    )
+    @mock.patch('subprocess.run')
     @mock.patch("yaml.dump")
     @mock.patch("os.path.exists", return_value=True)
     @mock.patch("os.remove", return_value=None)
@@ -626,19 +610,19 @@ def test_start_job_with_cli_args_gpu(
         mock_remove,
         mock_exists,
         mock_yaml_dump,
-        mock_file,
-        mock_main,
-        mock_compose,
-        mock_initialize_config_dir,
+        mock_subprocess_run,
     ):
         mock_validator = mock_validator_cls.return_value
         mock_validator.validate_aws_credential.return_value = True
         mock_kubernetes_client.get_current_context_namespace.return_value = "kubeflow"
         mock_get_console_link.return_value = "test-console-link"
         mock_yaml_dump.return_value = None
-        mock_main.return_value = None
-        mock_compose.return_value = None
-        mock_initialize_config_dir.return_value.__enter__.return_value = None
+        mock_subprocess_run.return_value = subprocess.CompletedProcess(
+            args=['some_command'],  # Simulate the command arguments
+            returncode=0,           # Simulate a successful command
+            stdout='Command executed successfully',  # Simulate standard output
+            stderr=''               # Simulate no errors
+        )
         result = self.runner.invoke(
             start_job,
             [
@@ -656,13 +640,7 @@ def test_start_job_with_cli_args_gpu(
         )
         self.assertEqual(result.exit_code, 0)
 
-    @mock.patch("hyperpod_cli.commands.job.initialize_config_dir")
-    @mock.patch("hyperpod_cli.commands.job.compose")
-    @mock.patch("hyperpod_cli.commands.job.customer_launcher")
-    @mock.patch(
-        "builtins.open",
-        new_callable=mock.mock_open,
-    )
+    @mock.patch('subprocess.run')
     @mock.patch("yaml.dump")
     @mock.patch("os.path.exists", return_value=True)
     @mock.patch("os.remove", return_value=None)
@@ -679,19 +657,19 @@ def test_start_job_with_cli_args_custom_label_selection(
         mock_remove,
         mock_exists,
         mock_yaml_dump,
-        mock_file,
-        mock_main,
-        mock_compose,
-        mock_initialize_config_dir,
+        mock_subprocess_run,
     ):
         mock_validator = mock_validator_cls.return_value
         mock_validator.validate_aws_credential.return_value = True
         mock_kubernetes_client.get_current_context_namespace.return_value = "kubeflow"
         mock_get_console_link.return_value = "test-console-link"
         mock_yaml_dump.return_value = None
-        mock_main.return_value = None
-        mock_compose.return_value = None
-        mock_initialize_config_dir.return_value.__enter__.return_value = None
+        mock_subprocess_run.return_value = subprocess.CompletedProcess(
+            args=['some_command'],  # Simulate the command arguments
+            returncode=0,           # Simulate a successful command
+            stdout='Command executed successfully',  # Simulate standard output
+            stderr=''               # Simulate no errors
+        )
         result = self.runner.invoke(
             start_job,
             [
@@ -711,10 +689,6 @@ def test_start_job_with_cli_args_custom_label_selection(
         )
         self.assertEqual(result.exit_code, 0)
 
-    @mock.patch(
-        "builtins.open",
-        new_callable=mock.mock_open,
-    )
     @mock.patch("yaml.dump")
     @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
     @mock.patch("hyperpod_cli.commands.job.JobValidator")
@@ -725,7 +699,6 @@ def test_start_job_with_cli_args_label_selection_not_json_str(
         mock_validator_cls,
         mock_kubernetes_client,
         mock_yaml_dump,
-        mock_file,
     ):
         mock_validator = mock_validator_cls.return_value
         mock_validator.validate_aws_credential.return_value = True
@@ -750,10 +723,6 @@ def test_start_job_with_cli_args_label_selection_not_json_str(
         )
         self.assertEqual(result.exit_code, 1)
 
-    @mock.patch(
-        "builtins.open",
-        new_callable=mock.mock_open,
-    )
     @mock.patch("yaml.dump")
     @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
     @mock.patch("hyperpod_cli.commands.job.JobValidator")
@@ -764,7 +733,6 @@ def test_start_job_with_cli_args_label_selection_invalid_values(
         mock_validator_cls,
         mock_kubernetes_client,
         mock_yaml_dump,
-        mock_file,
     ):
         mock_validator = mock_validator_cls.return_value
         mock_validator.validate_aws_credential.return_value = True
@@ -789,97 +757,87 @@ def test_start_job_with_cli_args_label_selection_invalid_values(
         )
         self.assertEqual(result.exit_code, 1)
 
-    @mock.patch("hyperpod_cli.commands.job.initialize_config_dir")
-    @mock.patch("hyperpod_cli.commands.job.compose")
-    @mock.patch("hyperpod_cli.commands.job.customer_launcher")
+    @mock.patch("hyperpod_cli.commands.job.validate_yaml_content")
+    @mock.patch("hyperpod_cli.commands.job.verify_and_load_yaml")
+    @mock.patch('subprocess.run')
+    @mock.patch("yaml.dump")
     @mock.patch("os.path.exists", return_value=True)
-    @mock.patch(
-        "builtins.open",
-        mock_open(read_data=VALID_CONFIG_FILE_DATA),
-    )
+    @mock.patch("os.remove", return_value=None)
     @mock.patch("hyperpod_cli.utils.get_cluster_console_url")
     @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
-    @mock.patch(
-        "os.path.abspath",
-        return_value="/absolute/path/to/file.yaml",
-    )
-    @mock.patch("os.path.isabs", return_value=False)
-    @mock.patch(
-        "os.path.split",
-        return_value=(
-            "/absolute/path/to",
-            "file.yaml",
-        ),
-    )
     @mock.patch("hyperpod_cli.commands.job.JobValidator")
     @mock.patch("boto3.Session")
     def test_start_job_with_config_file(
         self,
         mock_boto3,
         mock_validator_cls,
-        mock_split,
-        mock_isabs,
-        mock_abspath,
         mock_kubernetes_client,
         mock_get_console_link,
+        mock_remove,
         mock_exists,
-        mock_main,
-        mock_compose,
-        mock_initialize_config_dir,
+        mock_yaml_dump,
+        mock_subprocess_run,
+        mock_verify_and_load_yaml,
+        mock_validate_yaml_content,
     ):
         mock_validator = mock_validator_cls.return_value
         mock_validator.validate_aws_credential.return_value = True
+        mock_verify_and_load_yaml.return_value = yaml.safe_load(VALID_CONFIG_FILE_DATA)
+        mock_validate_yaml_content.return_value = True
+
         mock_kubernetes_client.get_current_context_namespace.return_value = "kubeflow"
         mock_get_console_link.return_value = "test-console-link"
-        mock_main.return_value = None
-        mock_compose.return_value = None
-        mock_initialize_config_dir.return_value.__enter__.return_value = None
+        mock_yaml_dump.return_value = None
+        mock_subprocess_run.return_value = subprocess.CompletedProcess(
+            args=['some_command'],  # Simulate the command arguments
+            returncode=0,           # Simulate a successful command
+            stdout='Command executed successfully',  # Simulate standard output
+            stderr=''               # Simulate no errors
+        )
         result = self.runner.invoke(
             start_job,
             ["--config-file", "file.yaml"],
+            catch_exceptions=False,
         )
+        print(result.exception)
         self.assertEqual(result.exit_code, 0)
 
-    @mock.patch("hyperpod_cli.commands.job.initialize_config_dir")
-    @mock.patch("hyperpod_cli.commands.job.compose")
-    @mock.patch("hyperpod_cli.commands.job.customer_launcher")
+    @mock.patch("hyperpod_cli.commands.job.validate_yaml_content")
+    @mock.patch("hyperpod_cli.commands.job.verify_and_load_yaml")
+    @mock.patch('subprocess.run')
+    @mock.patch("yaml.dump")
     @mock.patch("os.path.exists", return_value=True)
-    @mock.patch(
-        "builtins.open",
-        mock_open(read_data=VALID_CONFIG_FILE_DATA),
-    )
+    @mock.patch("os.remove", return_value=None)
     @mock.patch("hyperpod_cli.utils.get_cluster_console_url")
     @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
-    @mock.patch("os.path.isabs", return_value=True)
-    @mock.patch(
-        "os.path.split",
-        return_value=(
-            "/absolute/path/to",
-            "file.yaml",
-        ),
-    )
     @mock.patch("hyperpod_cli.commands.job.JobValidator")
     @mock.patch("boto3.Session")
     def test_start_job_with_config_file_absolute_path(
         self,
         mock_boto,
         mock_validator_cls,
-        mock_split,
-        mock_isabs,
         mock_kubernetes_client,
         mock_get_console_link,
+        mock_remove,
         mock_exists,
-        mock_main,
-        mock_compose,
-        mock_initialize_config_dir,
+        mock_dump,
+        mock_subprocess_run,
+        mock_verify_and_load_yaml,
+        mock_validate_yaml_content,
     ):
         mock_validator = mock_validator_cls.return_value
         mock_validator.validate_aws_credential.return_value = True
+        mock_verify_and_load_yaml.return_value = yaml.dump(VALID_CONFIG_FILE_DATA)
+        mock_validate_yaml_content.return_value = True
+
         mock_kubernetes_client.get_current_context_namespace.return_value = "kubeflow"
         mock_get_console_link.return_value = "test-console-link"
-        mock_main.return_value = None
-        mock_compose.return_value = None
-        mock_initialize_config_dir.return_value.__enter__.return_value = None
+        mock_subprocess_run.return_value = subprocess.CompletedProcess(
+            args=['some_command'],  # Simulate the command arguments
+            returncode=0,           # Simulate a successful command
+            stdout='Command executed successfully',  # Simulate standard output
+            stderr=''               # Simulate no errors
+        )
         result = self.runner.invoke(
             start_job,
             [
@@ -1017,14 +975,11 @@ def test_start_job_with_invalid_config_file(
         )
         self.assertNotEqual(result.exit_code, 0)
 
-    @mock.patch("hyperpod_cli.commands.job.initialize_config_dir")
-    @mock.patch("hyperpod_cli.commands.job.compose")
-    @mock.patch("hyperpod_cli.commands.job.customer_launcher")
-    @mock.patch(
-        "builtins.open",
-        new_callable=mock.mock_open,
-    )
+    @mock.patch("subprocess.run")
     @mock.patch("yaml.dump")
+    @mock.patch("os.path.exists", return_value=True)
+    @mock.patch("os.remove", return_value=None)
+    @mock.patch("hyperpod_cli.utils.get_cluster_console_url")
     @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
     @mock.patch("hyperpod_cli.commands.job.JobValidator")
     @mock.patch("boto3.Session")
@@ -1033,19 +988,22 @@ def test_start_job_with_cli_args_command_failed(
         mock_boto3,
         mock_validator_cls,
         mock_kubernetes_client,
+        mock_get_console_link,
+        mock_remove,
+        mock_exists,
         mock_yaml_dump,
-        mock_file,
-        mock_main,
-        mock_compose,
-        mock_initialize_config_dir,
+        mock_subprocess_run,
     ):
         mock_validator = mock_validator_cls.return_value
         mock_validator.validate_aws_credential.return_value = True
         mock_kubernetes_client.get_current_context_namespace.return_value = "kubeflow"
         mock_yaml_dump.return_value = None
-        mock_main.side_effect = Exception("submit job error")
-        mock_compose.return_value = None
-        mock_initialize_config_dir.return_value.__enter__.return_value = None
+        mock_subprocess_run.return_value = subprocess.CompletedProcess(
+            args=['some_command'],  # Simulate the command arguments
+            returncode=1,           # Simulate a failed command
+            stdout='Command failed',  # Simulate standard output
+            stderr=''               # Simulate no errors
+        )
         result = self.runner.invoke(
             start_job,
             [
@@ -1081,13 +1039,7 @@ def test_start_job_with_invalid_args(self, mock_validate):
         )
         self.assertEqual(result.exit_code, 1)
 
-    @mock.patch("hyperpod_cli.commands.job.initialize_config_dir")
-    @mock.patch("hyperpod_cli.commands.job.compose")
-    @mock.patch("hyperpod_cli.commands.job.customer_launcher")
-    @mock.patch(
-        "builtins.open",
-        new_callable=mock.mock_open,
-    )
+    @mock.patch("subprocess.run")
     @mock.patch("yaml.dump")
     @mock.patch("os.path.exists", return_value=True)
     @mock.patch("os.remove", return_value=None)
@@ -1104,19 +1056,19 @@ def test_start_job_with_cli_args_auto_resume_enabled(
         mock_remove,
         mock_exists,
         mock_yaml_dump,
-        mock_file,
-        mock_main,
-        mock_compose,
-        mock_initialize_config_dir,
+        mock_subprocess_run,
     ):
         mock_validator = mock_validator_cls.return_value
         mock_validator.validate_aws_credential.return_value = True
         mock_kubernetes_client.get_current_context_namespace.return_value = "kubeflow"
         mock_get_console_link.return_value = "test-console-link"
         mock_yaml_dump.return_value = None
-        mock_main.return_value = None
-        mock_compose.return_value = None
-        mock_initialize_config_dir.return_value.__enter__.return_value = None
+        mock_subprocess_run.return_value = subprocess.CompletedProcess(
+            args=['some_command'],  # Simulate the command arguments
+            returncode=0,           # Simulate a successful command
+            stdout='Command executed successfully',  # Simulate standard output
+            stderr=''               # Simulate no errors
+        )
         result = self.runner.invoke(
             start_job,
             [
@@ -1136,13 +1088,7 @@ def test_start_job_with_cli_args_auto_resume_enabled(
         )
         self.assertEqual(result.exit_code, 0)
 
-    @mock.patch("hyperpod_cli.commands.job.initialize_config_dir")
-    @mock.patch("hyperpod_cli.commands.job.compose")
-    @mock.patch("hyperpod_cli.commands.job.customer_launcher")
-    @mock.patch(
-        "builtins.open",
-        new_callable=mock.mock_open,
-    )
+    @mock.patch("subprocess.run")
     @mock.patch("yaml.dump")
     @mock.patch("os.path.exists", return_value=True)
     @mock.patch("os.remove", return_value=None)
@@ -1159,19 +1105,19 @@ def test_start_job_with_cli_args_deep_health_check_passed_nodes_only(
         mock_remove,
         mock_exists,
         mock_yaml_dump,
-        mock_file,
-        mock_main,
-        mock_compose,
-        mock_initialize_config_dir,
+        mock_subprocess_run,
     ):
         mock_validator = mock_validator_cls.return_value
         mock_validator.validate_aws_credential.return_value = True
         mock_kubernetes_client.get_current_context_namespace.return_value = "kubeflow"
         mock_get_console_link.return_value = "test-console-link"
         mock_yaml_dump.return_value = None
-        mock_main.return_value = None
-        mock_compose.return_value = None
-        mock_initialize_config_dir.return_value.__enter__.return_value = None
+        mock_subprocess_run.return_value = subprocess.CompletedProcess(
+            args=['some_command'],  # Simulate the command arguments
+            returncode=0,           # Simulate a successful command
+            stdout='Command executed successfully',  # Simulate standard output
+            stderr=''               # Simulate no errors
+        )
         result = self.runner.invoke(
             start_job,
             [
@@ -1191,13 +1137,7 @@ def test_start_job_with_cli_args_deep_health_check_passed_nodes_only(
         )
         self.assertEqual(result.exit_code, 0)
 
-    @mock.patch("hyperpod_cli.commands.job.initialize_config_dir")
-    @mock.patch("hyperpod_cli.commands.job.compose")
-    @mock.patch("hyperpod_cli.commands.job.customer_launcher")
-    @mock.patch(
-        "builtins.open",
-        new_callable=mock.mock_open,
-    )
+    @mock.patch("subprocess.run")
     @mock.patch("yaml.dump")
     @mock.patch("os.path.exists", return_value=True)
     @mock.patch("os.remove", return_value=None)
@@ -1214,19 +1154,19 @@ def test_start_job_with_cli_args_with_kueue(
         mock_remove,
         mock_exists,
         mock_yaml_dump,
-        mock_file,
-        mock_main,
-        mock_compose,
-        mock_initialize_config_dir,
+        mock_subprocess_run,
     ):
         mock_validator = mock_validator_cls.return_value
         mock_validator.validate_aws_credential.return_value = True
         mock_kubernetes_client.get_current_context_namespace.return_value = "kubeflow"
         mock_get_console_link.return_value = "test-console-link"
         mock_yaml_dump.return_value = None
-        mock_main.return_value = None
-        mock_compose.return_value = None
-        mock_initialize_config_dir.return_value.__enter__.return_value = None
+        mock_subprocess_run.return_value = subprocess.CompletedProcess(
+            args=['some_command'],  # Simulate the command arguments
+            returncode=0,           # Simulate a successful command
+            stdout='Command executed successfully',  # Simulate standard output
+            stderr=''               # Simulate no errors
+        )
         result = self.runner.invoke(
             start_job,
             [
@@ -1248,16 +1188,11 @@ def test_start_job_with_cli_args_with_kueue(
         )
         self.assertEqual(result.exit_code, 0)
 
-    @mock.patch("hyperpod_cli.commands.job.initialize_config_dir")
-    @mock.patch("hyperpod_cli.commands.job.compose")
-    @mock.patch("hyperpod_cli.commands.job.customer_launcher")
-    @mock.patch(
-        "builtins.open",
-        new_callable=mock.mock_open,
-    )
+    @mock.patch("subprocess.run")
     @mock.patch("yaml.dump")
     @mock.patch("os.path.exists", return_value=True)
     @mock.patch("os.remove", return_value=None)
+    @mock.patch("hyperpod_cli.utils.get_cluster_console_url")
     @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
     @mock.patch("hyperpod_cli.commands.job.JobValidator")
     @mock.patch("boto3.Session")
@@ -1266,21 +1201,24 @@ def test_start_job_with_cli_args_with_kueue_invalid(
         mock_boto3,
         mock_validator_cls,
         mock_kubernetes_client,
+        mock_get_console_link,
         mock_remove,
         mock_exists,
         mock_yaml_dump,
-        mock_file,
-        mock_main,
-        mock_compose,
-        mock_initialize_config_dir,
+        mock_subprocess_run,
     ):
         mock_validator = mock_validator_cls.return_value
         mock_validator.validate_aws_credential.return_value = True
+        mock_validator.validate_start_job_args.return_value = False
         mock_kubernetes_client.get_current_context_namespace.return_value = "kubeflow"
+        mock_get_console_link.return_value = "test-console-link"
         mock_yaml_dump.return_value = None
-        mock_main.return_value = None
-        mock_compose.return_value = None
-        mock_initialize_config_dir.return_value.__enter__.return_value = None
+        mock_subprocess_run.return_value = subprocess.CompletedProcess(
+            args=['some_command'],  # Simulate the command arguments
+            returncode=1,           # Simulate a failed command
+            stdout='Command failed',  # Simulate standard output
+            stderr=''               # Simulate no errors
+        )
         result = self.runner.invoke(
             start_job,
             [
@@ -1300,13 +1238,7 @@ def test_start_job_with_cli_args_with_kueue_invalid(
         )
         self.assertEqual(result.exit_code, 1)
 
-    @mock.patch("hyperpod_cli.commands.job.initialize_config_dir")
-    @mock.patch("hyperpod_cli.commands.job.compose")
-    @mock.patch("hyperpod_cli.commands.job.customer_launcher")
-    @mock.patch(
-        "builtins.open",
-        new_callable=mock.mock_open,
-    )
+    @mock.patch("subprocess.run")
     @mock.patch("yaml.dump")
     @mock.patch("os.path.exists", return_value=True)
     @mock.patch("os.remove", return_value=None)
@@ -1323,19 +1255,19 @@ def test_start_job_with_cli_args_with_service_account(
         mock_remove,
         mock_exists,
         mock_yaml_dump,
-        mock_file,
-        mock_main,
-        mock_compose,
-        mock_initialize_config_dir,
+        mock_subprocess_run,
     ):
         mock_validator = mock_validator_cls.return_value
         mock_validator.validate_aws_credential.return_value = True
         mock_kubernetes_client.get_current_context_namespace.return_value = "kubeflow"
         mock_get_console_link.return_value = "test-console-link"
         mock_yaml_dump.return_value = None
-        mock_main.return_value = None
-        mock_compose.return_value = None
-        mock_initialize_config_dir.return_value.__enter__.return_value = None
+        mock_subprocess_run.return_value = subprocess.CompletedProcess(
+            args=['some_command'],  # Simulate the command arguments
+            returncode=0,           # Simulate a successful command
+            stdout='Command executed successfully',  # Simulate standard output
+            stderr=''               # Simulate no errors
+        )
         result = self.runner.invoke(
             start_job,
             [
@@ -1355,13 +1287,7 @@ def test_start_job_with_cli_args_with_service_account(
         )
         self.assertEqual(result.exit_code, 0)
 
-    @mock.patch("hyperpod_cli.commands.job.initialize_config_dir")
-    @mock.patch("hyperpod_cli.commands.job.compose")
-    @mock.patch("hyperpod_cli.commands.job.customer_launcher")
-    @mock.patch(
-        "builtins.open",
-        new_callable=mock.mock_open,
-    )
+    @mock.patch("subprocess.run")
     @mock.patch("yaml.dump")
     @mock.patch("os.path.exists", return_value=True)
     @mock.patch("os.remove", return_value=None)
@@ -1378,19 +1304,19 @@ def test_start_job_with_cli_args_with_persistent_volume_claims(
         mock_remove,
         mock_exists,
         mock_yaml_dump,
-        mock_file,
-        mock_main,
-        mock_compose,
-        mock_initialize_config_dir,
+        mock_subprocess_run,
     ):
         mock_validator = mock_validator_cls.return_value
         mock_validator.validate_aws_credential.return_value = True
         mock_kubernetes_client.get_current_context_namespace.return_value = "kubeflow"
         mock_get_console_link.return_value = "test-console-link"
         mock_yaml_dump.return_value = None
-        mock_main.return_value = None
-        mock_compose.return_value = None
-        mock_initialize_config_dir.return_value.__enter__.return_value = None
+        mock_subprocess_run.return_value = subprocess.CompletedProcess(
+            args=['some_command'],  # Simulate the command arguments
+            returncode=0,           # Simulate a successful command
+            stdout='Command executed successfully',  # Simulate standard output
+            stderr=''               # Simulate no errors
+        )
         result = self.runner.invoke(
             start_job,
             [
@@ -1410,13 +1336,7 @@ def test_start_job_with_cli_args_with_persistent_volume_claims(
         )
         self.assertEqual(result.exit_code, 0)
 
-    @mock.patch("hyperpod_cli.commands.job.initialize_config_dir")
-    @mock.patch("hyperpod_cli.commands.job.compose")
-    @mock.patch("hyperpod_cli.commands.job.customer_launcher")
-    @mock.patch(
-        "builtins.open",
-        new_callable=mock.mock_open,
-    )
+    @mock.patch("subprocess.run")
     @mock.patch("yaml.dump")
     @mock.patch("os.path.exists", return_value=True)
     @mock.patch("os.remove", return_value=None)
@@ -1433,19 +1353,19 @@ def test_start_job_with_cli_args_with_local_volume(
         mock_remove,
         mock_exists,
         mock_yaml_dump,
-        mock_file,
-        mock_main,
-        mock_compose,
-        mock_initialize_config_dir,
+        mock_subprocess_run,
     ):
         mock_validator = mock_validator_cls.return_value
         mock_validator.validate_aws_credential.return_value = True
         mock_kubernetes_client.get_current_context_namespace.return_value = "kubeflow"
         mock_get_console_link.return_value = "test-console-link"
         mock_yaml_dump.return_value = None
-        mock_main.return_value = None
-        mock_compose.return_value = None
-        mock_initialize_config_dir.return_value.__enter__.return_value = None
+        mock_subprocess_run.return_value = subprocess.CompletedProcess(
+            args=['some_command'],  # Simulate the command arguments
+            returncode=0,           # Simulate a successful command
+            stdout='Command executed successfully',  # Simulate standard output
+            stderr=''               # Simulate no errors
+        )
         result = self.runner.invoke(
             start_job,
             [
@@ -1467,6 +1387,301 @@ def test_start_job_with_cli_args_with_local_volume(
         )
         self.assertEqual(result.exit_code, 0)
 
+    @mock.patch("subprocess.run")
+    @mock.patch("hyperpod_cli.utils.get_cluster_console_url")
+    @mock.patch("hyperpod_cli.service.discover_namespaces.DiscoverNamespaces.discover_accessible_namespace")
+    @mock.patch("yaml.dump")
+    @mock.patch("os.path.exists", return_value=True)
+    @mock.patch("os.remove", return_value=None)
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    @mock.patch("hyperpod_cli.commands.job.JobValidator")
+    @mock.patch("boto3.Session")
+    def test_start_job_with_cli_args_namespace_auto_discover(
+        self,
+        mock_boto3,
+        mock_validator_cls,
+        mock_kubernetes_client,
+        mock_remove,
+        mock_exists,
+        mock_yaml_dump,
+        mock_discover_accessible_namespace,
+        mock_get_cluster_console_url,
+        mock_subprocess_run,
+    ):
+        mock_validator = mock_validator_cls.return_value
+        mock_validator.validate_aws_credential.return_value = True
+        mock_kubernetes_client.get_current_context_namespace.return_value = None
+        mock_discover_accessible_namespace.return_value = "discovered_namespace"
+        mock_get_cluster_console_url.return_value = "test-console-link"
+
+        result = self.runner.invoke(
+            start_job,
+            [
+                "--job-name",
+                "test-job",
+                "--instance-type",
+                "ml.c5.xlarge",
+                "--image",
+                "pytorch:1.9.0-cuda11.1-cudnn8-runtime",
+                "--node-count",
+                "2",
+                "--entry-script",
+                "/opt/train/src/train.py",
+            ],
+            catch_exceptions=False
+        )
+
+        call_args_list = mock_yaml_dump.call_args_list
+        args, kwargs = call_args_list[0]
+
+        # Verify that namespace is filled with auto-discovered one
+        self.assertIn("'namespace': 'discovered_namespace'", str(args[0]))
+        self.assertEqual(result.exit_code, 0)
+    
+    @mock.patch('subprocess.run')
+    @mock.patch("hyperpod_cli.commands.job._get_auto_fill_queue_name")
+    @mock.patch("hyperpod_cli.commands.job.validate_yaml_content")
+    @mock.patch("hyperpod_cli.commands.job.verify_and_load_yaml")
+    @mock.patch("hyperpod_cli.utils.get_cluster_console_url")
+    @mock.patch("hyperpod_cli.service.discover_namespaces.DiscoverNamespaces.discover_accessible_namespace")
+    @mock.patch("yaml.dump")
+    @mock.patch("os.path.exists", return_value=True)
+    @mock.patch("os.remove", return_value=None)
+    @mock.patch("hyperpod_cli.utils.get_cluster_console_url")
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    @mock.patch("hyperpod_cli.commands.job.JobValidator")
+    @mock.patch("boto3.Session")
+    def test_start_job_with_cli_args_namespace_auto_fill_queue_name(
+        self,
+        mock_boto3,
+        mock_validator_cls,
+        mock_kubernetes_client,
+        mock_get_console_link,
+        mock_remove,
+        mock_exists,
+        mock_yaml_dump,
+        mock_discover_accessible_namespace,
+        mock_get_cluster_console_url,
+        mock_verify_and_load_yaml,
+        mock_validate_yaml_content,
+        mock_auto_fill_queue_name,
+        mock_subprocess_run,
+    ):
+        mock_validator = mock_validator_cls.return_value
+        mock_validator.validate_aws_credential.return_value = True
+        mock_discover_accessible_namespace.return_value = "test-namespace"
+        mock_get_cluster_console_url.return_value = "test-console-link"
+        mock_verify_and_load_yaml.return_value = yaml.safe_load(VALID_CONFIG_FILE_DATA)
+        mock_validate_yaml_content.return_value = True
+        mock_auto_fill_queue_name.return_value = "test-queue"
+
+        sm_managed_ns = V1Namespace(
+            metadata=V1ObjectMeta(
+                name="test-namespace",
+                labels={
+                    "sagemaker.amazonaws.com/sagemaker-managed-queue": "true",
+                    "sagemaker.amazonaws.com/quota-allocation-id": "test-team",
+                },
+            ),
+        )
+        mock_kubernetes_client().get_sagemaker_managed_namespace.return_value = sm_managed_ns
+
+        result = self.runner.invoke(
+            start_job,
+            [
+                "--job-name",
+                "test-job",
+                "--instance-type",
+                "ml.c5.xlarge",
+                "--image",
+                "pytorch:1.9.0-cuda11.1-cudnn8-runtime",
+                "--node-count",
+                "2",
+                "--entry-script",
+                "/opt/train/src/train.py",
+            ],
+            catch_exceptions=False
+        )
+
+        call_args_list = mock_yaml_dump.call_args_list
+        args, kwargs = call_args_list[0]
+
+        # Verify that the queue name is filled with SageMaker managed local queue
+        self.assertIn("'kueue.x-k8s.io/queue-name': 'test-queue'", str(args[0]))
+        self.assertEqual(result.exit_code, 0)
+
+    @mock.patch("subprocess.run")
+    @mock.patch("hyperpod_cli.utils.get_cluster_console_url")
+    @mock.patch("hyperpod_cli.service.discover_namespaces.DiscoverNamespaces.discover_accessible_namespace")
+    @mock.patch("yaml.dump")
+    @mock.patch("os.path.exists", return_value=True)
+    @mock.patch("os.remove", return_value=None)
+    @mock.patch("hyperpod_cli.utils.get_cluster_console_url")
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    @mock.patch("hyperpod_cli.commands.job.JobValidator")
+    @mock.patch("boto3.Session")
+    def test_start_job_with_cli_args_namespace_with_priority(
+        self,
+        mock_boto3,
+        mock_validator_cls,
+        mock_kubernetes_client,
+        mock_get_console_link,
+        mock_remove,
+        mock_exists,
+        mock_yaml_dump,
+        mock_discover_accessible_namespace,
+        mock_get_cluster_console_url,
+        mock_subprocess_run,
+    ):
+        mock_validator = mock_validator_cls.return_value
+        mock_validator.validate_aws_credential.return_value = True
+        mock_discover_accessible_namespace.return_value = "test-namespace"
+        mock_get_cluster_console_url.return_value = "test-console-link"
+        sm_managed_ns = V1Namespace(
+            metadata=V1ObjectMeta(
+                name="test-namespace",
+                labels={
+                    "sagemaker.amazonaws.com/sagemaker-managed-queue": "true",
+                    "sagemaker.amazonaws.com/quota-allocation-id": "test-team",
+                },
+            ),
+        )
+        mock_kubernetes_client().get_sagemaker_managed_namespace.return_value = sm_managed_ns
+
+        result = self.runner.invoke(
+            start_job,
+            [
+                "--job-name",
+                "test-job",
+                "--instance-type",
+                "ml.c5.xlarge",
+                "--image",
+                "pytorch:1.9.0-cuda11.1-cudnn8-runtime",
+                "--node-count",
+                "2",
+                "--entry-script",
+                "/opt/train/src/train.py",
+                "--priority",
+                "test-priority"
+            ],
+            catch_exceptions=False
+        )
+
+        call_args_list = mock_yaml_dump.call_args_list
+        args, kwargs = call_args_list[0]
+
+        # Verify that priority is passed correctly as a custom label
+        self.assertIn("'kueue.x-k8s.io/priority-class': 'test-priority'", str(args[0]))
+        self.assertEqual(result.exit_code, 0)
+
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    @mock.patch("hyperpod_cli.service.discover_namespaces.DiscoverNamespaces.discover_accessible_namespace")
+    def test_patch_job_with_namespace_success(self, mock_discover_accessible_namespace, mock_kubernetes_client):
+        mock_client = MagicMock()
+        mock_kubernetes_client.return_value = mock_client
+        mock_client.get_job.return_value = {
+            "metadata": {
+                "uid": "test-job-uid"
+            }
+        }
+        mock_client.get_workload_by_label.return_value = {
+            "items": [
+                {
+                    "metadata": {
+                        "name": "test-workload-name"
+                    }
+                }
+            ]
+        }
+        result = self.runner.invoke(
+            patch_job,
+            [
+                "suspend",
+                "--job-name",
+                "test-job",
+                "--namespace",
+                "test-namespace",
+            ],
+            catch_exceptions=False,
+        )
+
+        mock_client.patch_workload.assert_called_once()
+        call_args_list = mock_client.patch_workload.call_args_list
+        args, kwargs = call_args_list[0]
+
+        self.assertEqual(("test-workload-name", "test-namespace", {'spec': {'active': False}}), args)
+
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    @mock.patch("hyperpod_cli.service.discover_namespaces.DiscoverNamespaces.discover_accessible_namespace")
+    def test_patch_job_unsupported_patch_type(self, mock_discover_accessible_namespace, mock_kubernetes_client):
+        mock_client = MagicMock()
+        mock_kubernetes_client.return_value = mock_client
+        mock_client.get_job.return_value = {
+            "metadata": {
+                "uid": "test-job-uid"
+            }
+        }
+        mock_client.get_workload_by_label.return_value = {
+            "items": [
+                {
+                    "metadata": {
+                        "name": "test-workload-name"
+                    }
+                }
+            ]
+        }
+        result = self.runner.invoke(
+            patch_job,
+            [
+                "unsupported_type",
+                "--job-name",
+                "test-job",
+                "--namespace",
+                "test-namespace",
+            ],
+            catch_exceptions=False,
+        )
+
+        self.assertEqual(result.exit_code, 1)
+
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    @mock.patch("hyperpod_cli.service.discover_namespaces.DiscoverNamespaces.discover_accessible_namespace")
+    def test_patch_job_invalid_number_workloads(self, mock_discover_accessible_namespace, mock_kubernetes_client):
+        mock_client = MagicMock()
+        mock_kubernetes_client.return_value = mock_client
+        mock_client.get_job.return_value = {
+            "metadata": {
+                "uid": "test-job-uid"
+            }
+        }
+        mock_client.get_workload_by_label.return_value = {
+            "items": [
+                {
+                    "metadata": {
+                        "name": "test-workload-name-1"
+                    }
+                },
+                {
+                    "metadata": {
+                        "name": "test-workload-name-2"
+                    }
+                }
+            ]
+        }
+        result = self.runner.invoke(
+            patch_job,
+            [
+                "suspend",
+                "--job-name",
+                "test-job",
+                "--namespace",
+                "test-namespace",
+            ],
+            catch_exceptions=False,
+        )
+
+        self.assertEqual(result.exit_code, 1)
+
     def test_suppress_standard_output_context(
         self,
     ):
@@ -1522,3 +1737,182 @@ def test_config_file_with_other_arguments(self, mock_ctx):
         )
         with self.assertRaises(click.BadParameter):
             validate_only_config_file_argument(mock_ctx)
+
+    @mock.patch("subprocess.run")
+    @mock.patch("yaml.dump")
+    @mock.patch("os.path.exists", return_value=True)
+    @mock.patch("os.remove", return_value=None)
+    @mock.patch("hyperpod_cli.utils.get_cluster_console_url")
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    @mock.patch("hyperpod_cli.commands.job.JobValidator")
+    @mock.patch("boto3.Session")
+    def test_start_job_with_recipe(
+        self,
+        mock_boto3,
+        mock_validator_cls,
+        mock_kubernetes_client,
+        mock_get_console_link,
+        mock_remove,
+        mock_exists,
+        mock_yaml_dump,
+        mock_subprocess_run,
+    ):
+        mock_validator = mock_validator_cls.return_value
+        mock_validator.validate_aws_credential.return_value = True
+        mock_kubernetes_client.get_current_context_namespace.return_value = "kubeflow"
+        mock_get_console_link.return_value = "test-console-link"
+        mock_yaml_dump.return_value = None
+        mock_subprocess_run.return_value = subprocess.CompletedProcess(
+            args=['some_command'],  # Simulate the command arguments
+            returncode=0,           # Simulate a successful command
+            stdout='Command executed successfully',  # Simulate standard output
+            stderr=''               # Simulate no errors
+        )
+        result = self.runner.invoke(
+            start_job,
+            [
+                "--recipe",
+                "fine-tuning/llama/hf_llama3_8b_seq8192_gpu",
+            ],
+        )
+        self.assertEqual(result.exit_code, 0)
+
+    @mock.patch("subprocess.run")
+    @mock.patch("yaml.dump")
+    @mock.patch("os.path.exists", return_value=True)
+    @mock.patch("os.remove", return_value=None)
+    @mock.patch("hyperpod_cli.utils.get_cluster_console_url")
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    @mock.patch("hyperpod_cli.commands.job.JobValidator")
+    @mock.patch("boto3.Session")
+    def test_start_job_with_recipe_and_override_parameters(
+        self,
+        mock_boto3,
+        mock_validator_cls,
+        mock_kubernetes_client,
+        mock_get_console_link,
+        mock_remove,
+        mock_exists,
+        mock_yaml_dump,
+        mock_subprocess_run,
+    ):
+        mock_validator = mock_validator_cls.return_value
+        mock_validator.validate_aws_credential.return_value = True
+        mock_kubernetes_client.get_current_context_namespace.return_value = "kubeflow"
+        mock_get_console_link.return_value = "test-console-link"
+        mock_yaml_dump.return_value = None
+        mock_subprocess_run.return_value = subprocess.CompletedProcess(
+            args=['some_command'],  # Simulate the command arguments
+            returncode=0,           # Simulate a successful command
+            stdout='Command executed successfully',  # Simulate standard output
+            stderr=''               # Simulate no errors
+        )
+        override_params = '''{
+            "recipes.run.name": "test-run",
+            "recipes.trainer.num_nodes": 1,
+            "instance_type": "g5.48xlarge"
+        }'''
+
+        result = self.runner.invoke(
+            start_job,
+            [
+                "--recipe",
+                "fine-tuning/llama/hf_llama3_8b_seq8192_gpu",
+                "--override-parameters",
+                override_params,
+            ],
+        )
+        self.assertEqual(result.exit_code, 0)
+
+    @mock.patch("subprocess.run")
+    @mock.patch("yaml.dump")
+    @mock.patch("os.path.exists", return_value=True)
+    @mock.patch("os.remove", return_value=None)
+    @mock.patch("hyperpod_cli.utils.get_cluster_console_url")
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    @mock.patch("hyperpod_cli.commands.job.JobValidator")
+    @mock.patch("boto3.Session")
+    def test_start_job_with_recipe_invalid_override_parameters(
+        self,
+        mock_boto3,
+        mock_validator_cls,
+        mock_kubernetes_client,
+        mock_get_console_link,
+        mock_remove,
+        mock_exists,
+        mock_yaml_dump,
+        mock_subprocess_run,
+    ):
+        mock_validator = mock_validator_cls.return_value
+        mock_validator.validate_aws_credential.return_value = True
+        mock_kubernetes_client.get_current_context_namespace.return_value = "kubeflow"
+        mock_get_console_link.return_value = "test-console-link"
+        mock_yaml_dump.return_value = None
+        mock_subprocess_run.return_value = subprocess.CompletedProcess(
+            args=['some_command'],  # Simulate the command arguments
+            returncode=1,           # Simulate a failed command
+            stdout='Command failed',  # Simulate standard output
+            stderr=''               # Simulate no errors
+        )           
+        result = self.runner.invoke(
+            start_job,
+            [
+                "--recipe",
+                "fine-tuning/llama/hf_llama3_8b_seq8192_gpu",
+                "--override-parameters",
+                "invalid json",
+            ],
+        )
+        self.assertEqual(result.exit_code, 1)
+
+    @mock.patch("subprocess.run")
+    @mock.patch("yaml.dump")
+    @mock.patch("os.path.exists", return_value=True)
+    @mock.patch("os.remove", return_value=None)
+    @mock.patch("logging.Logger.debug")
+    @mock.patch("hyperpod_cli.utils.get_cluster_console_url")
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.__new__")
+    @mock.patch("hyperpod_cli.commands.job.JobValidator")
+    @mock.patch("boto3.Session")
+    def test_start_job_with_recipe_debug_mode(
+        self,
+        mock_boto3,
+        mock_validator_cls,
+        mock_kubernetes_client,
+        mock_get_console_link,
+        mock_debug,
+        mock_remove,
+        mock_exists,
+        mock_yaml_dump,
+        mock_subprocess_run,
+    ):
+        mock_validator = mock_validator_cls.return_value
+        mock_validator.validate_aws_credential.return_value = True
+        mock_kubernetes_client.get_current_context_namespace.return_value = "kubeflow"
+        mock_get_console_link.return_value = "test-console-link"
+        mock_yaml_dump.return_value = None
+        mock_subprocess_run.return_value = subprocess.CompletedProcess(
+            args=['some_command'],  # Simulate the command arguments
+            returncode=0,           # Simulate a successful command
+            stdout='Command executed successfully',  # Simulate standard output
+            stderr=''               # Simulate no errors
+        )
+        result = self.runner.invoke(
+            start_job,
+            [
+                "--recipe",
+                "fine-tuning/llama/hf_llama3_8b_seq8192_gpu",
+                "--debug",
+            ],
+        )
+        self.assertEqual(result.exit_code, 0)
+
+    def test_start_job_with_recipe_missing_required_args(self):
+        result = self.runner.invoke(
+            start_job,
+            [
+                "--override-parameters",
+                "{}",
+            ],
+        )
+        self.assertNotEqual(result.exit_code, 0)
diff --git a/test/unit_tests/validators/test_job_validator.py b/test/unit_tests/validators/test_job_validator.py
index 3b46802..253c614 100644
--- a/test/unit_tests/validators/test_job_validator.py
+++ b/test/unit_tests/validators/test_job_validator.py
@@ -11,10 +11,12 @@
 # ANY KIND, either express or implied. See the License for the specific
 # language governing permissions and limitations under the License.
 import unittest
-from unittest.mock import patch, mock_open
+from unittest.mock import patch, mock_open, MagicMock
+from unittest import mock
 
 from hyperpod_cli.constants.command_constants import (
     RestartPolicy,
+    SchedulerType,
 )
 from hyperpod_cli.validators.job_validator import (
     JobValidator,
@@ -47,7 +49,7 @@ def test_validate_start_job_args_job_valid(
             job_kind,
             command,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             None,
             False,
@@ -55,6 +57,7 @@ def test_validate_start_job_args_job_valid(
             None,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertTrue(result)
@@ -75,9 +78,8 @@ def test_validate_start_job_args_job_kind_invalid(self, mock_logger):
             instance_type,
             image,
             job_kind,
-            command,
-            None,
             None,
+            SchedulerType.KUEUE.value,
             None,
             None,
             False,
@@ -85,6 +87,7 @@ def test_validate_start_job_args_job_kind_invalid(self, mock_logger):
             None,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertFalse(result)
@@ -101,7 +104,6 @@ def test_validate_start_job_args_command_invalid(self, mock_logger):
         image = "image"
         job_kind = "kubeflow/PyTorchJob"
         command = "python"
-
         result = self.validator.validate_start_job_args(
             config_file,
             name,
@@ -111,7 +113,7 @@ def test_validate_start_job_args_command_invalid(self, mock_logger):
             job_kind,
             command,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             None,
             False,
@@ -119,6 +121,7 @@ def test_validate_start_job_args_command_invalid(self, mock_logger):
             None,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertFalse(result)
@@ -147,7 +150,7 @@ def test_validate_start_job_args_both_config_file_and_job_name_provided(
             job_kind,
             command,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             None,
             False,
@@ -155,11 +158,12 @@ def test_validate_start_job_args_both_config_file_and_job_name_provided(
             None,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertFalse(result)
         mock_logger.error.assert_called_once_with(
-            "Please provide only 'config-file' to submit job using config file or 'job-name' to submit job via CLI arguments"
+            "Please provide only 'config-file' to submit job using custom script or 'job-name' to submit job via CLI arguments"
         )
 
     @patch("hyperpod_cli.validators.job_validator.logger")
@@ -183,7 +187,7 @@ def test_validate_start_job_args_neither_config_file_nor_name_provided(
             job_kind,
             command,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             None,
             False,
@@ -191,11 +195,12 @@ def test_validate_start_job_args_neither_config_file_nor_name_provided(
             None,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertFalse(result)
         mock_logger.error.assert_called_once_with(
-            "Please provide either 'config-file' to submit job using config file or 'job-name' to submit job via CLI arguments"
+            "Please provide either 'recipe' for recipe-based jobs or 'config-file' to submit job using config file or 'job-name' to submit job via CLI arguments"
         )
 
     @patch("hyperpod_cli.validators.job_validator.logger")
@@ -219,7 +224,7 @@ def test_validate_start_job_args_name_provided_but_node_count_missing(
             job_kind,
             command,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             None,
             False,
@@ -227,6 +232,7 @@ def test_validate_start_job_args_name_provided_but_node_count_missing(
             None,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertFalse(result)
@@ -255,7 +261,7 @@ def test_validate_start_job_args_name_provided_but_instance_type_missing(
             job_kind,
             command,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             None,
             False,
@@ -263,6 +269,7 @@ def test_validate_start_job_args_name_provided_but_instance_type_missing(
             None,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertFalse(result)
@@ -291,7 +298,7 @@ def test_validate_start_job_args_name_provided_but_entry_script_missing(
             job_kind,
             command,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             None,
             False,
@@ -299,6 +306,7 @@ def test_validate_start_job_args_name_provided_but_entry_script_missing(
             None,
             "kubeflow",
             None,
+            None,
         )
 
         self.assertFalse(result)
@@ -328,7 +336,7 @@ def test_validate_start_job_args_auto_resume_in_namespace_with_aws_hyperpod_pref
             job_kind,
             command,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             None,
             auto_resume,
@@ -336,6 +344,7 @@ def test_validate_start_job_args_auto_resume_in_namespace_with_aws_hyperpod_pref
             max_retry,
             "hyperpod-1",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertTrue(result)
@@ -361,7 +370,7 @@ def test_validate_start_job_args_auto_resume_in_any_namespace(self, mock_logger)
             job_kind,
             command,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             None,
             auto_resume,
@@ -369,6 +378,7 @@ def test_validate_start_job_args_auto_resume_in_any_namespace(self, mock_logger)
             max_retry,
             "any-hyperpod",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertTrue(result)
@@ -393,7 +403,7 @@ def test_validate_start_job_args_no_auto_resume_has_max_retry(self, mock_logger)
             job_kind,
             command,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             None,
             None,
@@ -401,6 +411,7 @@ def test_validate_start_job_args_no_auto_resume_has_max_retry(self, mock_logger)
             max_retry,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertFalse(result)
@@ -431,7 +442,7 @@ def test_validate_start_job_args_auto_resume_wrong_restart_policy(
             job_kind,
             command,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             None,
             auto_resume,
@@ -439,6 +450,7 @@ def test_validate_start_job_args_auto_resume_wrong_restart_policy(
             max_retry,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertFalse(result)
@@ -467,7 +479,7 @@ def test_validate_start_job_args_name_provided_but_invalid_instance_type(
             job_kind,
             command,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             None,
             False,
@@ -475,6 +487,7 @@ def test_validate_start_job_args_name_provided_but_invalid_instance_type(
             None,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertFalse(result)
@@ -501,7 +514,7 @@ def test_validate_start_job_args_name_provided_but_image_missing(self, mock_logg
             job_kind,
             command,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             None,
             False,
@@ -509,6 +522,7 @@ def test_validate_start_job_args_name_provided_but_image_missing(self, mock_logg
             None,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertFalse(result)
@@ -538,13 +552,14 @@ def test_validate_start_job_args_valid_args(
             command,
             label_selector,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             False,
             None,
             None,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertTrue(result)
@@ -571,13 +586,14 @@ def test_validate_start_job_args_invalid_json_label_selector(
             command,
             label_selector,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             False,
             None,
             None,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertFalse(result)
@@ -604,13 +620,14 @@ def test_validate_start_job_args_invalid_values_label_selector(
             command,
             label_selector,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             False,
             None,
             None,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertFalse(result)
@@ -637,13 +654,14 @@ def test_validate_start_job_args_invalid_values_type_label_selector(
             command,
             label_selector,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             False,
             None,
             None,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertFalse(result)
@@ -670,13 +688,14 @@ def test_validate_start_job_args_invalid_key_label_selector(
             command,
             label_selector,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             False,
             None,
             None,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertFalse(result)
@@ -703,13 +722,14 @@ def test_validate_start_job_args_label_selector_unexpected_error(self, mock_json
             command,
             label_selector,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             False,
             None,
             None,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertFalse(result)
@@ -742,6 +762,7 @@ def test_validate_start_job_args_wrong_scheduler_type(
             None,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertFalse(result)
@@ -774,9 +795,10 @@ def test_validate_start_job_args_kueue_fields_error(
             None,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
-        self.assertFalse(result)
+        self.assertTrue(result)
 
     def test_validate_start_job_args_auto_resume_restart_policy_unexpected(
         self,
@@ -798,7 +820,7 @@ def test_validate_start_job_args_auto_resume_restart_policy_unexpected(
             job_kind,
             command,
             None,
-            None,
+            SchedulerType.KUEUE.value,
             None,
             None,
             True,
@@ -806,50 +828,216 @@ def test_validate_start_job_args_auto_resume_restart_policy_unexpected(
             1,
             "kubeflow",
             "/opt/train/src/train.py",
+            None,
         )
 
         self.assertFalse(result)
 
-    @patch("os.path.exists", return_value=False)
-    def test_validate_start_job_config_yaml_file_not_exists(self, mock_exists):
-        result = self.validator.validate_start_job_config_yaml("non_existing_file.yaml")
-        self.assertFalse(result)
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.get_sagemaker_managed_namespace")
+    def test_validate_start_job_args_sagemaker_scheduler_success(
+        self,
+        mock_get_sagemaker_managed_namespace,
+    ):
+        config_file = None
+        name = "job-name"
+        node_count = "2"
+        instance_type = "ml.p4d.24xlarge"
+        image = "my-image:latest"
+        job_kind = "kubeflow/PyTorchJob"
+        command = "torchrun"
 
-    @patch("os.path.exists", return_value=True)
-    @patch(
-        "hyperpod_cli.validators.job_validator.verify_and_load_yaml",
-        return_value=None,
-    )
-    def test_validate_start_job_config_yaml_invalid_yaml(
+        mock_get_sagemaker_managed_namespace.return_value = "kubeflow"
+
+        result = self.validator.validate_start_job_args(
+            config_file,
+            name,
+            node_count,
+            instance_type,
+            image,
+            job_kind,
+            command,
+            None,
+            SchedulerType.SAGEMAKER.value,
+            None,
+            None,
+            False,
+            None,
+            None,
+            "kubeflow",
+            "/opt/train/src/train.py",
+        )
+
+        self.assertTrue(result)
+
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.get_sagemaker_managed_namespace")
+    def test_validate_start_job_args_sagemaker_scheduler_invalid_namespace(
         self,
-        mock_verify_and_load_yaml,
-        mock_exists,
+        mock_get_sagemaker_managed_namespace,
     ):
-        result = self.validator.validate_start_job_config_yaml("test.yaml")
+        config_file = None
+        name = "job-name"
+        node_count = "2"
+        instance_type = "ml.p4d.24xlarge"
+        image = "my-image:latest"
+        job_kind = "kubeflow/PyTorchJob"
+        command = "torchrun"
+
+        mock_get_sagemaker_managed_namespace.return_value = None
+
+        result = self.validator.validate_start_job_args(
+            config_file,
+            name,
+            node_count,
+            instance_type,
+            image,
+            job_kind,
+            command,
+            None,
+            SchedulerType.SAGEMAKER.value,
+            None,
+            None,
+            False,
+            None,
+            None,
+            "kubeflow",
+            "/opt/train/src/train.py",
+        )
+
         self.assertFalse(result)
 
-    @patch("os.path.exists", return_value=True)
-    @patch("hyperpod_cli.validators.job_validator.verify_and_load_yaml")
-    @patch(
-        "hyperpod_cli.validators.job_validator.validate_yaml_content",
-        return_value=True,
-    )
-    def test_validate_start_job_config_yaml_valid(
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.list_workload_priority_classes")
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.get_sagemaker_managed_namespace")
+    def test_validate_start_job_args_sagemaker_scheduler_success_with_priority(
         self,
-        mock_validate_yaml_content,
-        mock_verify_and_load_yaml,
-        mock_exists,
+        mock_get_sagemaker_managed_namespace,
+        mock_list_workload_priority_classes,
     ):
-        mock_data = {
-            "cluster": {
-                "cluster_type": "k8s",
-                "cluster_config": {},
-            }
+        config_file = None
+        name = "job-name"
+        node_count = "2"
+        instance_type = "ml.p4d.24xlarge"
+        image = "my-image:latest"
+        job_kind = "kubeflow/PyTorchJob"
+        command = "torchrun"
+
+        mock_get_sagemaker_managed_namespace.return_value = "kubeflow"
+        mock_list_workload_priority_classes.return_value = {
+            'items': [
+                {'metadata': {'name': 'high-priority'}},
+                {'metadata': {'name': 'medium-priority'}},
+                {'metadata': {'name': 'test-priority'}}
+            ]
         }
-        mock_verify_and_load_yaml.return_value = mock_data
-        result = self.validator.validate_start_job_config_yaml("test.yaml")
+
+        result = self.validator.validate_start_job_args(
+            config_file,
+            name,
+            node_count,
+            instance_type,
+            image,
+            job_kind,
+            command,
+            None,
+            SchedulerType.SAGEMAKER.value,
+            None,
+            "test-priority",
+            False,
+            None,
+            None,
+            "kubeflow",
+            "/opt/train/src/train.py",
+        )
+
         self.assertTrue(result)
 
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.list_workload_priority_classes")
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.get_sagemaker_managed_namespace")
+    def test_validate_start_job_args_sagemaker_scheduler_fail_priority_not_exist(
+        self,
+        mock_get_sagemaker_managed_namespace,
+        mock_list_workload_priority_classes,
+    ):
+        config_file = None
+        name = "job-name"
+        node_count = "2"
+        instance_type = "ml.p4d.24xlarge"
+        image = "my-image:latest"
+        job_kind = "kubeflow/PyTorchJob"
+        command = "torchrun"
+
+        mock_get_sagemaker_managed_namespace.return_value = "kubeflow"
+        mock_list_workload_priority_classes.return_value = {
+            'items': [
+                {'metadata': {'name': 'high-priority'}},
+                {'metadata': {'name': 'medium-priority'}},
+            ]
+        }
+
+        result = self.validator.validate_start_job_args(
+            config_file,
+            name,
+            node_count,
+            instance_type,
+            image,
+            job_kind,
+            command,
+            None,
+            SchedulerType.SAGEMAKER.value,
+            None,
+            "test-priority",
+            False,
+            None,
+            None,
+            "kubeflow",
+            "/opt/train/src/train.py",
+        )
+
+        self.assertFalse(result)
+    
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.list_workload_priority_classes")
+    @mock.patch("hyperpod_cli.clients.kubernetes_client.KubernetesClient.get_sagemaker_managed_namespace")
+    def test_validate_start_job_args_sagemaker_scheduler_no_namespace(
+        self,
+        mock_get_sagemaker_managed_namespace,
+        mock_list_workload_priority_classes,
+    ):
+        config_file = None
+        name = "job-name"
+        node_count = "2"
+        instance_type = "ml.p4d.24xlarge"
+        image = "my-image:latest"
+        job_kind = "kubeflow/PyTorchJob"
+        command = "torchrun"
+
+        mock_get_sagemaker_managed_namespace.return_value = "kubeflow"
+        mock_list_workload_priority_classes.return_value = {
+            'items': [
+                {'metadata': {'name': 'high-priority'}},
+                {'metadata': {'name': 'medium-priority'}},
+            ]
+        }
+
+        result = self.validator.validate_start_job_args(
+            config_file,
+            name,
+            node_count,
+            instance_type,
+            image,
+            job_kind,
+            command,
+            None,
+            SchedulerType.SAGEMAKER.value,
+            None,
+            "test-priority",
+            False,
+            None,
+            None,
+            None,
+            "/opt/train/src/train.py",
+        )
+
+        self.assertFalse(result)
+
     @patch("os.path.exists", return_value=True)
     @patch(
         "builtins.open",
@@ -875,6 +1063,15 @@ def test_verify_and_load_yaml_valid(self, mock_exists):
     def test_verify_and_load_yaml_invalid(self, mock_exists):
         result = verify_and_load_yaml("test.yaml")
         self.assertIsNone(result)
+    
+    @patch("os.path.exists", return_value=False)
+    @patch(
+        "builtins.open",
+        mock_open(read_data="cluster:\n  cluster_type: k8s\n  cluster_config: {}"),
+    )
+    def test_verify_and_load_yaml_not_exist(self, mock_exists):
+        result = verify_and_load_yaml("test.yaml")
+        self.assertFalse(result)
 
     def test_validate_yaml_content_valid(self):
         mock_data = {
@@ -943,8 +1140,13 @@ def test_validate_yaml_content_valid_with_queue_name_and_priority(
         result = validate_yaml_content(mock_data)
         self.assertTrue(result)
 
+    @patch(
+        "hyperpod_cli.validators.job_validator.validate_scheduler_related_fields",
+        return_value=True,
+    )
     def test_validate_yaml_content_valid_with_auto_resume(
         self,
+        mock_validate_scheduler_related_fields,
     ):
         mock_data = {
             "cluster": {
@@ -964,8 +1166,13 @@ def test_validate_yaml_content_valid_with_auto_resume(
         result = validate_yaml_content(mock_data)
         self.assertTrue(result)
 
+    @patch(
+        "hyperpod_cli.validators.job_validator.validate_scheduler_related_fields",
+        return_value=True,
+    )
     def test_validate_yaml_content_valid_with_auto_resume_in_aws_hyperpod_namespace(
         self,
+        mock_validate_scheduler_related_fields,
     ):
         mock_data = {
             "cluster": {
@@ -985,7 +1192,14 @@ def test_validate_yaml_content_valid_with_auto_resume_in_aws_hyperpod_namespace(
         result = validate_yaml_content(mock_data)
         self.assertTrue(result)
 
-    def test_validate_yaml_content_valid_with_auto_resume_default_namespace(self):
+    @patch(
+        "hyperpod_cli.validators.job_validator.validate_scheduler_related_fields",
+        return_value=True,
+    )
+    def test_validate_yaml_content_valid_with_auto_resume_default_namespace(    
+        self,
+        mock_validate_scheduler_related_fields,
+    ):
         mock_data = {
             "cluster": {
                 "cluster_type": "k8s",
@@ -1004,8 +1218,13 @@ def test_validate_yaml_content_valid_with_auto_resume_default_namespace(self):
         result = validate_yaml_content(mock_data)
         self.assertTrue(result)
 
+    @patch(
+        "hyperpod_cli.validators.job_validator.validate_scheduler_related_fields",
+        return_value=True,
+    )
     def test_validate_yaml_content_valid_with_auto_resume_any_hyperpod_namespace(
         self,
+        mock_validate_scheduler_related_fields,
     ):
         mock_data = {
             "cluster": {
@@ -1025,8 +1244,13 @@ def test_validate_yaml_content_valid_with_auto_resume_any_hyperpod_namespace(
         result = validate_yaml_content(mock_data)
         self.assertTrue(result)
 
+    @patch(
+        "hyperpod_cli.validators.job_validator.validate_scheduler_related_fields",
+        return_value=True,
+    )
     def test_validate_yaml_content_error_only_auto_resume_no_max_retry(
         self,
+        mock_validate_scheduler_related_fields,
     ):
         mock_data = {
             "cluster": {
@@ -1045,8 +1269,13 @@ def test_validate_yaml_content_error_only_auto_resume_no_max_retry(
         result = validate_yaml_content(mock_data)
         self.assertFalse(result)
 
+    @patch(
+        "hyperpod_cli.validators.job_validator.validate_scheduler_related_fields",
+        return_value=True,
+    )
     def test_validate_yaml_content_error_with_wrong_restart_policy(
         self,
+        mock_validate_scheduler_related_fields,
     ):
         mock_data = {
             "cluster": {
@@ -1066,8 +1295,13 @@ def test_validate_yaml_content_error_with_wrong_restart_policy(
         result = validate_yaml_content(mock_data)
         self.assertFalse(result)
 
+    @patch(
+        "hyperpod_cli.validators.job_validator.validate_scheduler_related_fields",
+        return_value=True,
+    )
     def test_validate_yaml_content_error_only_max_retry(
         self,
+        mock_validate_scheduler_related_fields,
     ):
         mock_data = {
             "cluster": {
@@ -1142,20 +1376,6 @@ def test_validate_hyperpod_related_fields_invalid_max_retry(
         )
         self.assertFalse(result)
 
-    def test_validate_hyperpod_related_fields_invalid_queue_and_priority(
-        self,
-    ):
-        result = validate_hyperpod_related_fields(
-            instance_type="ml.g5.xlarge",
-            queue_name=None,
-            priority="high-priority",
-            auto_resume=False,
-            restart_policy=RestartPolicy.NEVER.value,
-            max_retry=None,
-            namespace="kubeflow",
-        )
-        self.assertFalse(result)
-
 
-if __name__ == "__main__":
+if __name__== "__main__":
     unittest.main()