Power up your ML workflows with interactive IDEs on SageMaker HyperPod

This article explains how to use Amazon SageMaker Spaces, a new feature enabling interactive development environments like JupyterLab and VS Code on SageMaker HyperPod clusters with EKS orchestration.

• SageMaker Spaces provides managed JupyterLab and Code Editor environments for data scientists
• Supports fractional GPU allocations via NVIDIA MIG technology for cost efficiency
• Quick install option automatically configures IAM roles, remote access, and EKS add-ons
• Data scientists create Spaces using HyperPod CLI or kubectl commands
• Access Spaces via web UI with custom domain or remote VS Code connection
• Supports private and public Spaces with real-time collaboration capabilities
• Automatic idle shutdown optimizes resource usage by stopping inactive workspaces
• Integrates with HyperPod task governance and observability for resource management
• Administrators can create custom templates with pre-configured settings for teams

SageMaker Spaces streamlines ML development by providing secure, managed environments on HyperPod clusters, reducing setup complexity and enabling teams to focus on model development.