MLflow

MLflow tracks the parameters, metrics, and artifacts of every ML training run so a team can compare experiments and reproduce results, packages trained models in a standard format that works across serving frameworks, and manages a model registry for versioning and promoting models toward production. It's framework-agnostic, working with PyTorch, TensorFlow, scikit-learn, and others already in this catalog.

MLflow is a fit for data science teams that need to track experiments and manage models systematically rather than in scattered notebooks and spreadsheets, filling an MLOps gap that's currently absent from this catalog's ML framework entries.