SYSTEM Cited by 1 source

Amazon SageMaker AI Pipelines

Amazon SageMaker AI Pipelines is AWS's managed ML-workflow orchestration service within the SageMaker AI family. Pipelines are defined as a directed graph of steps (data prep, training, evaluation, registration) with first-class ML concepts (model registry, conditions, caching, parameters).

Stub — expand as dedicated sources arrive.

Seen in

• sources/2026-04-01-aws-automate-safety-monitoring-with-computer-vision-and-generative-ai — Canonical 7-step pipeline shape: checkpoint loading → data preparation + split → training → drift-baseline generation → evaluation → model packaging → model registration. Data scientists trigger pipelines with flexible hyperparameter + ground-truth selection scripts. Model approval on the resulting registered model fires an EventBridge event that triggers a model-promotion Lambda to open a code review against the application-infrastructure repo, cleanly decoupling science (approve when metrics meet criteria) from application (merge + deploy via CI/CD).