Developer Pods

Developer Pods provide serverless, on-demand GPU compute for development, training, and inference workloads. You choose a pod from the catalog, configure it, deploy it, and access it over SSH.

---

Step 1: Open Developer Pods Catalog

Open the Developer Pods catalog to view available SKUs, GPU types, and hourly pricing.

The catalog includes options such as:

• 1 x H100 GPU
• 2 x H100 GPU
• 1 x L40S GPU
• 2 x L40S GPU

Select the SKU that matches your workload profile.

---

Step 2: Configure Pod Basics

After selecting a catalog item, enter the pod configuration values:

• Name (required)
• Description (optional)
• Compute SKU (auto-selected from catalog)
• Workspace (required)

---

Step 3: Configure Compute and Runtime

Set the compute profile and image/runtime options for the pod.

Choose the pod image from the available options (for example Ubuntu and PyTorch/CUDA variants).

---

Step 4: Review Cost Estimate

Before deployment, review the cost panel showing estimated monthly and hourly pricing for the selected configuration.

Info

Pods are billed hourly, and minimum usage can be a fraction of an hour.

---

Step 5: View Deployed Pods

After deploying, use the pods list to monitor status and manage instances across workspaces.

The list view provides:

• Pod name
• Workspace
• Created timestamp
• Status
• Actions menu

---

Step 6: Access Pod via SSH

Open a pod from the list to view its details and output section.

Use the provided commands in order:

1. Create/download the private key file.
2. Run the SSH command to connect to the pod.

Example format:

ssh -i <key-file>.key ubuntu@<public-ip> -p <port>

---

Step 7: Delete Pod

From the actions menu in the pods list, choose Delete to remove a pod.

Warning

Pod deletion is irreversible. Back up any required data before deleting.

---