Using Amazon SageMaker for governance
Throughout this chapter, we have detailed the importance of establishing an AI governance framework. However, setting up an overall process to gain visibility of performance, control access, audit changes, and mitigate bias is no easy feat. To help address these challenges and remove undifferentiated heavy lifting, SageMaker provides purpose-built tools to help implement governance.
ML governance capabilities with Amazon SageMaker
As user adoption increases, it becomes more difficult for administrators to manage user access to ML projects. Custom permissions policies are often required for different ML user groups and these permissions sets vary greatly. Customization is a time-consuming process that could delay user onboarding. Amazon SageMaker Role Manager (https://docs.aws.amazon.com/sagemaker/latest/dg/role-manager.html) simplifies this process by providing a baseline set of permissions for different user personas and ML activities through...