Orchestrate ML Workflow using Amazon SageMaker Pipelines and Model Registry

Why it matters

1.2.7 governs custom model artifacts through Model Registry versions, evaluation evidence, and approval states; this Lab builds the pipeline that actually produces those versions.

What to practise

Assemble a SageMaker Pipeline that trains or evaluates a model and registers the result in Model Registry, then inspect the registered version and its approval state.

How it fits

This is the production step behind 1.2.7’s registry decisions, and it feeds the promotion-and- rollback choice in 1.2.8: a version only has somewhere to be promoted from because a pipeline like this one registered it.