A Model Registry in MLOps is a centralized system used to store, track, and manage machine learning models throughout their lifecycle. It helps teams organize different model versions, maintain metadata, and control the promotion of models from development to production. This improves collaboration, reproducibility, and governance in machine learning workflows. A Model Registry also makes it easier to monitor, update, and roll back models when necessary. In your opinion, how important is a Model Registry for managing ML models at scale, and what problems can arise if teams do not use one?