Run interactive IDEs on Amazon EKS with SageMaker AI to power up your AI workflows

This article explains how to deploy the SageMaker AI Spaces add-on on Amazon EKS to run interactive IDEs like JupyterLab and Code Editor directly on your Kubernetes cluster.

• Consolidates interactive and training workloads on one cluster, improving GPU utilization by up to 30 percent
• Reduces setup time from 3–5 days to approximately 5 minutes for data scientists to launch a fully configured Space
• Provides browser access via presigned URLs and VS Code remote access through SSH-over-SSM tunnels
• Requires AWS Load Balancer Controller, ACM certificate, KMS key for JWT encryption, and proper IAM roles with Pod Identity
• Supports OIDC sign-in with Amazon Cognito for corporate credential authentication and self-service workspace management
• Eliminates need for separate JupyterHub deployments while maintaining access to GPU nodes, shared storage, and IAM roles

By running interactive IDEs on the same EKS cluster as training pipelines, teams reduce operational complexity and accelerate time-to-productivity for machine learning development.