Lewis Tunstall
Machine Learning Engineer at Hugging Face
Bern, Bern, Switzerland
Summary
🤩
Rockstar🎓
Top SchoolLewis Tunstall is a Machine Learning Engineer with seven years of experience building production-ready ML tooling and educational content, currently on the research team at Hugging Face in Bern. He specializes in ML infrastructure and deployment—pushing models and model cards to the Hugging Face Hub, improving ONNX runtime integrations, and enabling RLHF workflows—while also contributing widely to datasets, transformers, and evaluation libraries. Lewis blends research rigor (PhD-level background in theoretical physics) with practical MLOps and full-stack notebook work, having helped ship few-shot, diffusion, and speech-pretraining tooling used across the community. He also mentors AI startups at HoistAI and has a track record of translating complex ML topics into accessible course material and multilingual documentation. A not-obvious strength: his background in theoretical physics underpins a methodical approach to model evaluation and data pipeline design, helping bridge academic ideas and reliable engineering.
7 years of coding experience
8 years of employment as a software developer
Doctor of Philosophy (PhD), Theoretical and Mathematical Physics, Doctor of Philosophy (PhD), Theoretical and Mathematical Physics at University of Adelaide
English, Spanish, German
Stackoverflow
1reputation
47reached
0answers
1question
Github Skills (65)
transformers10
hub10
python10
testing10
automatic-speech-recognition10
exporter10
datasets10
huggingface10
deep-learning10
trainings10
few-shot-learning10
machine-translation10
scikit10
internationalization10
data-export10
Programming languages (8)
Github contributions (5)
huggingface/open-r1
Jan 2025 - Jul 2025
Fully open reproduction of DeepSeek-R1
Role in this project:
ML Engineer
ML EngineerContributions:133 reviews, 130 PRs, 322 pushes in 6 months
Contributions summary:Lewis primarily focused on setting up and refining data loading and processing functionalities within the `open-r1` repository, likely for the DeepSeek-R1 model. They added core components like the `data.py` file, implemented dataset mixing logic, and integrated chat template application, which suggests work on data preparation for supervised fine-tuning (SFT) tasks. The contributions include refactoring the evaluation procedures and adding custom evaluation tasks using LightEval for model performance assessment. Furthermore, the commits also show updates to training scripts and configurations, indicating the user's involvement in the model's training and evaluation pipeline.
huggingface/course
Jan 2022 - Jan 2023
The Hugging Face course on Transformers
Role in this project:Technical Writer & Localization Specialist
Technical Writer & Localization SpecialistContributions:399 reviews, 266 commits, 413 PRs in 11 months
Contributions summary:Lewis primarily contributed to the course by translating and adapting content for different languages. Their commits include fixing formatting errors, correcting translation issues, and completing the translation of chapters. Furthermore, the user's work involved adding missing sections and adapting the course structure for multilingual support, demonstrating a focus on expanding the course's accessibility. They also focused on content organization and adapting the course structure for multilingual support.
huggingfacedeep-learningnlptransformers
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.