Amazon SageMaker Feature Store now supports cross-account sharing, discovery, and access

This article discusses the new cross-account sharing, discovery, and access capabilities of Amazon SageMaker Feature Store, which allows users to share feature groups across different AWS accounts.

Specifically, the article covers:

• The need for a centralized feature store with cross-account access for collaboration and governance in machine learning projects
• Delineating the roles of data producers (data/ML engineers) and data consumers (data scientists) in a cross-account setup
• The types of permissions that can be granted: discoverability permissions (view feature group names and metadata) and access permissions (read-only, read/write, or admin access to feature groups)
• Step-by-step guide on how to grant and accept discoverability and access permissions for feature groups across accounts using AWS Resource Access Manager (RAM)
• Sample notebooks demonstrating the cross-account sharing and access workflows
• Benefits of cross-account feature sharing, including collaboration, governance, security, and efficiency in model development