DVC

DVC, short for Data Version Control, brings Git-style versioning to machine learning datasets, models, and experiments. Developed by Iterative and released under the Apache 2.0 license, it stores large artifacts in cloud or on-premise storage such as S3, Azure, Google Cloud, or SSH remotes while keeping small metadata files under regular Git control, so a repository can describe exactly which data and model versions produced a given result. DVC also defines reproducible pipelines as computational graphs connecting code and data, re-running only the stages affected by a change, and tracks experiments locally against existing Git hosting without extra servers. Comparison commands let teams diff data, parameters, metrics, and models across runs. The project ships as a command-line tool plus a VS Code extension and is widely used by data scientists and ML engineers who want reproducibility and collaboration built on workflows they already know from Git.