ML platform
Amazon SageMaker
Managed platform for data preparation, training, tuning, deployment, and monitoring of ML models.
Key points
- Provides managed capabilities across the ML lifecycle: prepare data, build, train, tune, deploy, and monitor.
- Supports notebooks, training jobs, hosting endpoints, pipelines, feature management, and model governance features.
- Works for custom ML when teams need more control than prebuilt AI services provide.
- Can train models using built-in algorithms, custom code, and containerized environments.
- Supports MLOps patterns such as repeatable pipelines, model registry, monitoring, and CI/CD integration.
When to use it
- Choose SageMaker when a team must build, train, tune, and deploy custom ML models.
- Use it when prebuilt AI services do not fit the business problem or data requirements.
- Use it for production MLOps with model monitoring, pipelines, and governed deployment.
Exam tips
- SageMaker is the ML platform; Bedrock is the managed foundation model platform for generative AI.
- Use prebuilt AI services such as Comprehend or Rekognition when the task is standard and no custom model is needed.
- Autopilot, Canvas, JumpStart, Pipelines, Clarify, and Model Monitor are SageMaker capabilities with distinct exam clues.
- SageMaker requires ML lifecycle choices; AI services are simpler managed APIs for specific tasks.