MLFlow in general focuses more exclusively on the "lineage tracking" pieces alone. I.e. registering your models and such. They do have an experimental pipeline product as you mention, but it's different in a number of ways. I'd say the biggest one is that MLFlow pipelines has fixed, pre-determined structures for your pipelines while Sematic lets you build your own. We also have a more python-dev-friendly API and better access to cloud computing resources. You can find more details on our comparison page: https://docs.sematic.dev/sematic-vs#...-mlflow-pipelines