Grégoire Martinon
AI Research Director at Emerton
Greater Paris Metropolitan Region France
Summary
🤩
RockstarGrégoire Martinon is an AI research director based in Greater Paris with five years of focused experience building reliable, trustworthy AI for industry and defense. He has led R&D programs in AI robustness, transparency, and ethical algorithms at firms like Capgemini Invent and Quantmetry, translating research into production-ready practices and documentation for high-risk AI projects. His background in numerical relativity and physics gives him deep expertise in numerical methods, HPC and rigorous modeling—skills he leverages to design uncertainty-aware systems and lifecycle metrology for models. An active contributor to scikit-learn-contrib/MAPIE, he implemented core conformal prediction features and testing that strengthen prediction-interval tooling for classification and regression. Colleagues know him for combining scientific rigor with pragmatic delivery: building agentic translation tools for defense, explainability packages for finance and cosmetics, and coaching teams on LLM adoption. He’s as comfortable writing unit-tested Python libraries as he is shaping governance and sustainability metrics for AI in production.
5 years of coding experience
8 years of employment as a software developer
French, English, Italian
Github Skills (9)
scikit-learn10
regression10
machine-learning10
pytest10
python10
data-science10
classification10
numpy10
scikit10
Programming languages (4)
Github contributions (5)
scikit-learn-contrib/MAPIE
Apr 2021 - Apr 2022
A scikit-learn-compatible library for estimating prediction intervals and controlling risks, based on conformal predictions.
Role in this project:
Data Scientist
Data ScientistContributions:1 release, 337 reviews, 366 commits in 1 year
Contributions summary:Grégoire's contributions focused on implementing and testing core functionality for the `mapie` library, specifically related to prediction intervals and conformity scores. They added unit tests for metrics, implemented new methods for calculating confidence intervals and expanded the scope of the library with new features, such as the "prefit" method. The user made code improvements by refactoring parts of the codebase, including the addition of the `sample_weight` parameter. The user's work directly supports the library's goal of estimating prediction sets for classification and regression problems.
regressionpythonconfidence-intervalspredictiondata-science
simai-ml/how-to-opensource
Jun 2021 - Jul 2021
"How to opensource" BBL material
Contributions:11 reviews, 68 commits, 9 PRs in 14 days
bbl
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.