Model Registry
Domain: MLOps
🎯 Motivation
A centralized store for trained models: versions, metadata, performance metrics. Like a library for models.
📋 Concrete Example
Registry: {model: sentiment-v1, accuracy: 88%, date: 2025-01-01, status: production}. {model: sentiment-v2, accuracy: 91%, date: 2025-03-01, status: staging}. All versions tracked.
🔗 Analogy
A library catalog - each book has metadata (title, author, date, location). A model registry catalogs models with their metadata and versions.