Thomas Moreau
Chargé De Recherche at Inria
Paris, Ile-de-France
Summary
🤩
Rockstar🎓
Top SchoolThomas Moreau is a research scientist at Inria Saclay with 12 years of experience specializing in machine learning, signal processing and high-dimensional statistics, particularly applied to physiological signals. He completed a PhD on convolutional sparse representations and now focuses on convolutional dictionary learning, learned optimization algorithms (e.g., LISTA) and their theoretical and computational properties. Thomas blends academic rigor with production-grade engineering: he is an active open-source contributor to major Python projects such as joblib, loky and CPython where his fixes improved robustness of parallel processing and reduced test-suite and memory issues in scikit-learn and simulation-based inference tooling. Based in Paris, he combines algorithmic innovation with practical systems work—optimizing process pools, addressing pickling/deadlock bugs and scaling parallel computation for scientific workloads. Less obvious: his research ties interpretability of convolutional models to concrete signal-analysis applications (gait, oculomotor recordings), bridging theory, code and real-world physiological insights.
12 years of coding experience
M.Sc and ingenieur diploma Computer science and Applied Mathematics, M.Sc and ingenieur diploma Computer science and Applied Mathematics at École PolytechniqueDiplôme d'ingénieur Mathématiques et informatique, Diplôme d'ingénieur Mathématiques et informatique at Télécom ParisMathematics and Physics, Mathematics and Physics at Lycée ChampollionDoctorate Mathématiques appliquées, Doctorate Mathématiques appliquées at École Normale Supérieure Paris-Saclay
English, French
Stackoverflow
4,437reputation
358kreached
64answers
0questions
python-multiprocessing
top-5%
multiprocessing
top-5%
tensorflow
top-5%
python
top-5%
Github Skills (30)
pytorch10
debug10
pytest10
python10
multithreading10
testing10
scikit10
machine-learning10
python-multiprocessing10
parallel-computing10
hyper-threading10
thread10
scikit-learn10
multiprocessing10
multi-process10
Programming languages (12)
Github contributions (5)
joblib/joblib
Aug 2017 - Sep 2022
Computing with Python functions.
Role in this project:
Back-end Developer
Back-end DeveloperContributions:5 releases, 232 reviews, 28 commits in 5 years 2 months
Contributions summary:Thomas contributed significantly to the `joblib/joblib` repository, primarily focusing on backend improvements and enhancements. Their work involved integrating the `loky` library as the default backend, optimizing process management, and addressing pickling-related issues for functions and objects defined in the `__main__` module. They also improved threadpool management, and contributed to code maintenance. These changes improved the stability and usability of `joblib`'s parallel processing capabilities.
pythonmemoizationcachingmultiprocessingcomputing
scikit-learn/scikit-learn
Oct 2015 - Oct 2019
scikit-learn: machine learning in Python
Role in this project:ML Engineer
ML EngineerContributions:1 review, 15 commits, 7 PRs in 4 years 1 month
Contributions summary:Thomas primarily contributed to the `scikit-learn` library by enhancing clustering metric functionalities and optimizing test execution times. They introduced a `max_n_classes` parameter to several clustering metrics to handle potential memory errors, specifically within `contingency_matrix`, `adjusted_rand_score`, `homogeneity_completeness_v_measure`, `homogeneity_score`, `completeness_score`, `v_measure_score`, `mutual_info_score`, `adjusted_mutual_info_score`, and `normalized_mutual_info_score`. Additionally, they corrected typos and improved the test suite to reduce overall test durations.
data-analysispythonstatisticsdata-sciencelearn-machine-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.