Repolode
Submit
zenml-io

zenml-io/zenml

Open-source Python platform for building ML and AI workflows as pipelines that run on your own or cloud infrastructure, from training models to agents.

View on GitHubWebsite
Stars
5,622
Forks
665
Open issues
172
Last commit
20 hours ago

What is it?

ZenML is a Python framework for ML and AI engineers who need to turn experiments into repeatable, production-grade workflows. You write pipelines as ordinary Python functions and steps, such as training a model or running an agent loop, and ZenML runs them on a configurable infrastructure backend called a stack.

For each run, ZenML containerizes the code, tracks metrics, logs, and metadata, and abstracts away the underlying infrastructure. It integrates with existing tools including MLflow, LangGraph, Langfuse, SageMaker, and Vertex AI, so you can keep your current libraries and services rather than replacing them.

ZenML uses a client-server architecture with a web dashboard. It can run fully locally for development, or connect to a separately deployed server for production use.

Key features

  • 01Write pipelines and steps as plain Python functions.
  • 02Automatically containerizes and tracks code, runs, metrics, and artifacts.
  • 03Runs on configurable stacks across local and cloud infrastructure.
  • 04Integrates with MLflow, Weights & Biases, LangGraph, Langfuse, SageMaker, and Vertex AI.
  • 05Provides a web dashboard and a VS Code / Cursor extension for managing pipelines.

Use cases

  • Moving classical ML training and batch inference from notebooks to repeatable pipelines.
  • Building and evaluating RAG or LLM applications with tracked runs.
  • Deploying ML models or agents as production endpoints with versioning and monitoring.
  • Standardizing MLOps across a team that uses several existing ML tools.

Quick start

  1. 1

    Install ZenML with server capabilities

    pip install "zenml[server]"
  2. 2

    Initialize your ZenML repository

    zenml init
  3. 3

    Start a local server or connect to a remote one

    zenml login

Things to know

  • Server-side features need the larger zenml[server] install; plain pip install zenml gives only a slimmer client.
  • Many integrations (cloud, MLflow, Kubernetes and others) require their own setup and infrastructure.

Topics

#mlops#machine-learning#llm#pipelines#workflow-orchestration#agents#python#model-deployment#agentops#ai

FAQ

What is zenml?
Open-source Python platform for building ML and AI workflows as pipelines that run on your own or cloud infrastructure, from training models to agents. ZenML is a Python framework for ML and AI engineers who need to turn experiments into repeatable, production-grade workflows. You write pipelines as ordinary Python functions and steps, such as training a model or running an agent loop, and ZenML runs them on a configurable infrastructure backend called a stack.
Is zenml free to use? What license does it have?
zenml is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is zenml still maintained?
Actively maintained. The last commit to zenml was on Oct 21, 2026.
What can I use zenml for?
Moving classical ML training and batch inference from notebooks to repeatable pipelines. Building and evaluating RAG or LLM applications with tracked runs. Deploying ML models or agents as production endpoints with versioning and monitoring. Standardizing MLOps across a team that uses several existing ML tools.
How do I install zenml?
Install ZenML with server capabilities: pip install "zenml[server]" Initialize your ZenML repository: zenml init Start a local server or connect to a remote one: zenml login
Where can I get zenml?
The source code and installation instructions are on GitHub: https://github.com/zenml-io/zenml

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects