TPOT is an open-source Python AutoML tool that automatically designs and optimizes machine learning pipelines using genetic programming. Developed by the Epistasis Lab, it has gathered more than 10,000 stars on GitHub and remains actively maintained, with updates in 2025. Think of it as a data science assistant that searches thousands of pipeline combinations so you do not have to hand-tune them.
TPOT stands for Tree-based Pipeline Optimization Tool. Given a dataset, it uses genetic programming to evolve pipelines made of preprocessing steps, feature engineering, model selection, and hyperparameter settings, scoring each candidate with cross-validation. The project recently went through a major refactoring: the package was rewritten from scratch for better efficiency, a more modular framework, and easier customization of the evolutionary process. Newer capabilities include genetic feature selection, multi-objective optimization that can balance accuracy against pipeline complexity, and a significantly expanded way of defining search spaces. TPOT is built on scikit-learn, so the pipelines it discovers export as familiar scikit-learn code you can inspect, edit, and ship. It is distributed through PyPI and conda-forge.
TPOT is free and open source under the LGPL-3.0 license. There are no subscriptions, seat fees, or hosted upsells: install it locally, run it on your own machines, and keep the pipelines it finds. Costs are limited to your own compute time during the search.
TPOT fits data scientists and analysts who want a strong baseline fast, researchers in fields like biomedical informatics who need reproducible model search, and ML engineers who prefer generated scikit-learn code over black-box AutoML services. It is especially useful when you have tabular data and limited time for manual tuning.