Benjamin Bossan
Machine Learning Engineer at Hugging Face
Berlin, Germany
Summary
🤩
Rockstar🎓
Top SchoolBenjamin Bossan is a Machine Learning Engineer with 11 years of experience building and leading data science teams, currently contributing to Hugging Face from Berlin. He combines deep academic rigor—a Ph.D. in Biology—with hands-on ML and Python engineering, shipping production-ready features and CI/CD improvements across major open-source projects like Transformers and Accelerate. As a former Head of Data Science and Lead ML Engineer, he has moved teams from research to production, specializing in efficient fine-tuning (PEFT), mixed-precision training, and robust checkpointing. His open-source work spans model interoperability, test coverage and DevOps automation, reflecting a blend of research-minded model interpretability (visualization work on nolearn) and pragmatic backend stability. Notably, he has solved tricky model-pickling and adapter-loading edge cases that improve reproducibility for large-scale PyTorch workflows.
11 years of coding experience
13 years of employment as a software developer
Baccalauréat, Abitur, Baccalauréat, Abitur at Lycée Français de Berlin/Französisches Gymnasium BerlinDiplom, Biology, General, Diplom, Biology, General at Humboldt-Universität zu Berlin
German, English, French, Latin
Stackoverflow
21reputation
177reached
1answer
0questions
Github Skills (32)
transformers10
unit-testing10
pytorch10
convolutional-neural-networks10
acc10
pytest10
python10
matplotlib10
scikit10
testing10
acceleration10
machine-learning10
accelerator10
lasagne10
multiprecision10
Programming languages (9)
Github contributions (5)
huggingface/peft
Jun 2023 - Apr 2025
🤗 PEFT: State-of-the-art Parameter-Efficient Fine-Tuning.
Role in this project:
Back-end Developer & DevOps Engineer
Back-end Developer & DevOps EngineerContributions:9 releases, 1328 reviews, 769 PRs in 1 year 9 months
Contributions summary:Benjamin primarily focused on improving the test coverage and stability of the project. They removed unnecessary dependencies, added testing frameworks for reporting coverage, and fixed various bugs. Their contributions involved refactoring existing code and adding checks to ensure the project's reliability. They also appear to have worked on build and deployment, which may be related to fixing issues related to the continuous integration/continuous delivery pipeline.
parametermachine-learningfineparameter-efficientfine-tuning
skorch-dev/skorch
Sep 2017 - Jan 2023
A scikit-learn compatible neural network library that wraps PyTorch
Role in this project:ML Engineer
ML EngineerContributions:9 releases, 231 reviews, 201 commits in 5 years 5 months
Contributions summary:Benjamin made several commits to the skorch library, which is a scikit-learn compatible neural network library that wraps PyTorch. The commits involved refactoring existing code, adding new features such as support for torch.compile, and making the binary classifier work with BCELoss. They also made improvements to the documentation and added testing for various features including the GPyTorch integration.
pytorchpythonwrapsneural-networksmachine-learning
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.