François Chollet
Co-Founder at Ndea
United States
Summary
🤩
RockstarFrançois Chollet is a machine learning engineer and entrepreneur with 11 years of experience building and maintaining core deep learning infrastructure, now co-founding Ndea and ARC Prize. He spent nearly a decade at Google driving Keras and TensorFlow integration, model implementations, documentation, and tooling—work that helped shape widely used projects like Keras, TensorFlow Probability, and popular model zoos (ResNet, Inception, Xception). François combines research-grade understanding of probabilistic modeling and dimensionality reduction with hands-on engineering: refactoring Stable Diffusion in TensorFlow, adding Keras 3 support to libraries like ParametricUMAP, and building hyperparameter tuning primitives. He’s known for improving code quality and testability across large open-source repos, and for translating ML concepts into clear educational notebooks and docs tied to his "Deep Learning with Python" materials. Less obvious: he often focuses on maintainability and API hygiene—removing deprecated APIs, standardizing tests, and making components backend-agnostic—which amplifies long-term impact across the ML ecosystem.
11 years of coding experience
9 years of employment as a software developer
Github Skills (43)
convolutional-neural-networks10
probabilistic-programming10
pytest10
python10
dimensionality-reduction10
imap10
image-processing10
image-classification10
testing10
machine-learning10
stable-diffusion10
recurrent-neural-networks10
numpy10
hyperparameter-optimization10
keras10
Programming languages (9)
Github contributions (5)
fchollet/deep-learning-with-python-notebooks
Sep 2017 - Feb 2022
Role in this project:
ML Engineer
ML EngineerContributions:5 reviews, 28 commits, 13 PRs in 4 years 5 months
Contributions summary:François contributed extensively to the provided deep-learning notebooks. Their work included adding and modifying code samples, specifically in the context of Keras and TensorFlow, to align with the content of the "Deep Learning with Python" book. The changes encompass various areas of deep learning, showcasing their knowledge of practical implementation, including binary classification and model training. The user appears to have experience in adapting and updating the codebase for newer versions of TensorFlow.
pytorchpythondeep-learningjupyter-notebookmachine-learning
keras-team/keras-preprocessing
May 2018 - Feb 2022
Utilities for working with image data, text data, and sequence data.
Role in this project:Back-end Developer
Back-end DeveloperContributions:6 releases, 1 review, 42 commits in 3 years 9 months
Contributions summary:François primarily focused on the development and maintenance of the Keras preprocessing library. They implemented and refined image processing functionalities, including image data augmentation, and added features such as directory iteration. Their contributions involved fixing code style issues, adding testing modules for core features, and improving overall code maintainability and error handling within the library. This included ensuring compatibility with different Keras backends.
text-datasequencepythonimage-data
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.