BYOI: Custom Container Environments

Bring Your Own Image (BYOI) documentation for Amazon SageMaker AI—enabling enterprise teams to use custom container environments in AWS's AI/ML platform.

Role: Primary Author
Audience: ML Engineers, Data Scientists, Enterprise Administrators
Type: How-to Guide, Reference Documentation

View the Live Documentation

Note on Live Documentation

As with all actively maintained documentation, these pages may have been updated by other contributors since original publication. The core structure and approach I established remain the foundation of these documentation nodes.

The Challenge

Enterprise administrators and ML engineers often need to standardize their development environments across teams, like specific Python versions, pre-installed libraries, security configurations. Before this documentation existed, there was no comprehensive guide for bringing custom Docker images into SageMaker AI Studio or SageMaker Unified Studio.

I tested code and authored the complete documentation node from scratch for both platforms, covering everything from image specifications to advanced optimization techniques.

My Approach

  • Hands-on testing: I worked through every workflow myself, often with a more basic setup than the engineering team, which helped me identify edge cases and common pitfalls users would encounter.
  • Engineering collaboration: Found issues in the original workflow examples and proposed solutions. Engineering and I iterated together to create the most efficient, error-free how-to guides.
  • Cross-service strategy: Coordinated with the SageMaker Unified Studio team to handle overlapping content while serving each platform's unique requirements.
  • AI-assisted validation: Used AI assistance to troubleshoot issues in workflow examples, finding solutions that weren't immediately obvious from engineering's reference environment.

Documentation Structure

I designed an information architecture that guides users from basic concepts to advanced optimization:

  • Bring your own image (BYOI) — Overview and use cases
    • Image specifications — Technical requirements
    • How to BYOI — Step-by-step workflows
      • Create a custom image and push to Amazon ECR
      • Attach your custom image to your domain
      • Update container configuration
    • Launch a custom image in Studio
    • View your custom image details
    • Speed up container startup with SOCI — Advanced optimization
      • Permissions for SOCI indexing
      • Create SOCI indexes with nerdctl and SOCI CLI example
      • Integrate SOCI-indexed images with Studio example
    • Detach and clean up custom image resources

What This Demonstrates

Technical depth in AI/ML containerization workflows, ability to structure complex how-to content, developer empathy (testing from a user's perspective), and effective cross-team collaboration with engineering.