AWS - Machine Learning Engineer Associate Study Guide

4: ML Solution Monitoring, Maintenance, and Security

This chapter closes the loop on the MLA-C01 workflow by watching production inference for drift and quality regressions, holding infrastructure inside cost and latency envelopes, and locking access to models, data, and pipelines with identity, network, and audit controls. It covers SageMaker Model Monitor, Clarify online explainability, A/B and shadow testing, CloudWatch and CloudTrail signals, cost management with Cost Explorer, Budgets, Trusted Advisor, and Savings Plans, right-sizing with Inference Recommender and Compute Optimizer, and least-privilege IAM plus VPC isolation for SageMaker resources.