ZenML

Open source MLOps framework (Apache 2.0) that turns your ML code into versioned, reproducible pipelines instead of throwaway scripts. You write your steps in Python, ZenML tracks every artifact, model and run in one metadata store, and the same pipeline runs locally for debugging then on Kubernetes or in the cloud without changing your logic. It covers classical ML (scikit-learn, PyTorch) as well as LLM and agent pipelines (LangGraph, LlamaIndex), with 60+ integrations to wire in your existing tools.

Strengths

Limitations

Best for

Official site

View on Coeurdar