Key Building Blocks
- Processing jobs transform and validate data before training.
- Training jobs run built-in algorithms or custom containers.
- Model Registry stores versions, metadata, approval status, and deployment readiness.
- Endpoints host models for real-time, serverless, or asynchronous inference.
Exam Cues
- Need repeatable ML workflow: SageMaker Pipelines.
- Need systematic hyperparameter search: automatic model tuning.
- Need detect drift after deployment: SageMaker Model Monitor.
- Need cost-optimized interruptible training: managed spot training.
Practice SageMaker Questions
Put your knowledge to the test with practice questions.