Home · Frameworks · DVC
FRAMEWORK

DVC

Git-based data version control and ML pipeline management tool for reproducible experiments.

TARGET QUERY dvc data version control · ~8K/mo
CREATOR
Iterative
LANGUAGE
Python
CATEGORY
MLOps
GITHUB STARS
~14K
LICENSE
Apache 2.0
OVERVIEW Updated 2026-05-17

What It Does

DVC (Data Version Control) extends Git to handle large files, datasets, and ML pipelines. It versions data alongside code, tracks experiment pipelines, and enables reproducible ML workflows without storing large files in Git repositories.

Key Features

DVC offers large file versioning with remote storage backends (S3, GCS, Azure), pipeline definition and execution, experiment tracking and comparison, metrics visualization, model registry, and seamless Git integration for team collaboration.

When To Use It

Use DVC when you need to version large datasets alongside code, create reproducible ML pipelines, or when your team already uses Git and wants data versioning that fits that workflow. It excels at making data science projects reproducible.

Alternatives

MLflow provides broader lifecycle management. Git LFS handles large files but lacks pipeline features. Pachyderm offers Kubernetes-native data versioning. Weights & Biases focuses more on experiment tracking than data versioning.