Samuel Humeau
AI Scientist at Mistral AI
Paris, Ile-de-France
Summary
🤩
Rockstar🎓
Top SchoolSamuel Humeau is an AI Scientist with 11 years of experience building large-scale machine learning systems that extract and organize information from the web. He has progressed from research internships to roles at Diffbot and Facebook and led ML teams at Nabla before joining Mistral AI, blending deep research rigor with production-focused engineering. His work includes backend and model contributions to high-profile open-source projects like Facebook Research’s ParlAI, where he implemented and optimized bi-encoder and cross-encoder components. Trained at École Polytechnique and EPFL (MSc in Machine Learning), he combines strong theoretical foundations with practical skills in search, transcription, translation, and recommendation systems. Known for squeezing performance and compatibility gains out of complex models, he focuses on turning vast noisy web data into the kind of structured knowledge bases that power smarter decision-making. Based in Paris, he brings a rare mix of academic excellence, industry impact, and hands-on contributions to foundational dialogue and extraction toolkits.
11 years of coding experience
7 years of employment as a software developer
Preparation for Ecole Polytechnique competition entrance exam Mathematics and physics, Preparation for Ecole Polytechnique competition entrance exam Mathematics and physics at Lycée ClemenceauDiplome de l'école polytechnique Electrical engineering, Diplome de l'école polytechnique Electrical engineering at École PolytechniqueMaster of Science (M.Sc.) Computer Science Machine Learning 5.54/6.0, Master of Science (M.Sc.) Computer Science Machine Learning 5.54/6.0 at EPFL
French, English, German
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (12)
pytorch10
machine-learning10
nlp10
trainings10
python10
natural-language-processing10
modeling10
evaluation9
eval9
deep-learning9
transformer-models9
backend8
Programming languages (7)
Github contributions (5)
facebookresearch/ParlAI
Dec 2018 - Jul 2019
A framework for training and evaluating AI models on a variety of openly available dialogue datasets.
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:68 commits, 41 PRs, 122 pushes in 7 months
Contributions summary:Samuel primarily contributed to the development of the bi-encoder and cross-encoder models within the ParlAI framework. Their work involved implementing and refining core components of these models, including loss functions, candidate scoring mechanisms, and optimizations. The commits also included updates to model initialization, performance enhancements, and compatibility adjustments within the existing framework.
nlpdeep-learningdatasetmachine-learningtraining
Cerbal/frohmd
May 2016 - Jun 2017
Contributions:13 pushes, 1 branch in 1 year 1 month
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.