Definition
A model registry is a centralized repository that stores trained machine learning models along with their metadata and manages each model through its lifecycle. Think of it as version control for models: every trained artifact gets a unique version identifier, and models move through defined stages — development, staging, production, and archived — with a clear record of who promoted what and when.
What the registry stores
Beyond the model weights themselves, a registry captures the metadata that makes a model reproducible and auditable: the training parameters, evaluation metrics, the dataset version it was trained on, and lineage back to the experiment that produced it. The large binary artifacts usually live in scalable object storage, while the registry tracks the pointers, stages, and approvals. Tools like MLflow, Weights & Biases, and SageMaker Model Registry implement this pattern.
Governance and reproducibility
The registry is the backbone of governance in MLOps. It answers questions that matter when a model misbehaves in production: which exact version is live, what data it was trained on, and how to roll back to a known-good version. This is the same reproducibility discipline a sovereign operator wants over any critical system — a full, auditable history rather than an opaque artifact someone handed you.
From the registry, an approved model is pushed to model serving and then watched with model monitoring. Versioning is what makes safe rollback and rollout possible.
In Simple Terms
A model registry is a centralized repository that stores trained machine learning models along with their metadata and manages each model through its lifecycle. Think…