Question 37
Domain 2 — AI Operations, Lifecycle, and Control EnvironmentIn MLOps, which component stores ALL versions of deployed models along with their metadata, training lineage, and lifecycle stage?
Correct answer: D
Explanation
A model registry is the central system for managing model artifacts and their governance. It stores "all versions of deployed models" together with metadata, training lineage, and lifecycle stage so teams can track, approve, and promote models through development, staging, and production.
Why each option is right or wrong
A. Feature store — manages feature definitions for training and serving
B. Data catalog — documents dataset metadata and lineage
C. Experiment tracker — logs training runs and hyperparameters
D. Model registry — central store for model artifacts with lifecycle governance
A model registry is the MLOps system of record for model artifacts, and it is the component that tracks each registered version together with metadata, training lineage, and stage transitions such as development, staging, and production. By contrast, model serving or deployment tools only host a live model endpoint and do not maintain the full version history or governance state required to manage multiple releases.