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

The Amazon SageMaker AI Spaces add-on for Amazon EKS allows data scientists to run interactive IDEs such as JupyterLab and Code Editor directly on their existing Kubernetes clusters. This integration eliminates the need to switch to standalone environments, which often lack the GPU nodes, shared storage, and IAM roles that are essential for complex AI workflows.

Setting up the add-on involves installing and configuring it on an Amazon EKS cluster, deploying supporting components like the AWS Load Balancer Controller, and establishing secure connections via TLS certificates and IAM roles. Once configured, users can access their Spaces through browser-based URLs or from VS Code over SSH-over-SSM, enabling seamless development and collaboration.

The architecture leverages AWS services such as Route 53 for DNS resolution, ACM for TLS certificates, and KMS for encryption keys, ensuring secure and reliable access. The solution also supports moving to OpenID Connect sign-in with Amazon Cognito, allowing teams to authenticate using corporate credentials.

This approach consolidates AI development environments onto a single cluster, improving GPU utilization and reducing costs associated with dedicated notebook fleets. The process requires specific prerequisites, including an AWS account, CLI tools, and a Route 53 domain, and involves creating roles, policies, and deploying add-ons through Helm and AWS CLI commands.

The post provides comprehensive steps for deploying the add-on, creating access entries, and managing resources, along with cleanup instructions to avoid ongoing charges. Overall, this setup enhances AI workflows by providing quick, secure, and integrated development environments directly on existing Kubernetes infrastructure.