contact firstname.lastname [@] gmail [DOT] com
GitHub, Twitter, BlueSky

I am a tenured researcher (Chargé de recherche) in Inria Lyon, in the OCKHAM team, working in Machine Learning. I am also a part-time lecturer (Professeur chargé de cours) at Ecole Polytechnique (Python for datasciences, statistics), and I teach in ENS de Lyon (machine and deep learning, optimization, optimal transport, generative models).

I have a keen interest in the Python programming language: I am the lead developer of skglm (fast and flexible sklearn GLMs) and celer (fastest Lasso solver). I am also a core developer of benchopt, a benchmarking framework that makes optimization benchmarks easy, transparent and reproducible.

I am an Area Chair for NeurIPS and ICML, and an associate editor for TMLR; I was previously involved as associated editor for Computo, a journal promoting reproducibility through a novel publication format.

I co-organize the Machine Learning and Signal Processing seminar at ENS Lyon, send me a message if you want to present your work there!

I wrote a series of posts summarizing advices on permanent positions at CNRS/INRIA, avoiding pitfalls in scientific presentations and finding undergraduate internships.

Short bio: From January 2020 to October 2021, I was a post-doc at University of Genova, where I worked with Lorenzo Rosasco and Silvia Villa to develop new algorithms for implicit regularization. I hold a PhD from Télécom Paris and Inria Saclay (Parietal Team), under the supervision of Joseph Salmon and Alexandre Gramfort. In my PhD, I improved the efficiency of brain signals reconstruction algorithms (more details here), which involved optimization, sparsity and high dimensional statistics. I received the PGMO PhD prize for this work. You can find more details on my résumé (12/2025) and my list of publications.

Job offers

• None currently available, but contact me if interested.

People

Current team:

• Georges Le Bellier, Postdoc/RI, since May 2026.
• Anne Gagneux, PhD student from Ponts/MVA co-supervised with Emmanuel Soubies and Rémi Gribonval, since April 2023 (defense Dec. 2026).
• Orel Mazor, M1 intern from Mines Paris, co-supervised with Quentin Bertrand and Rémi Emonet, since January 2026.
• Alexandre Lagier, M2 intern from ENS Lyon, co-supervised with Ségolène Martin and Anne Gagneux, since February 2026.
• Ishan Nath, M2 intern from ENS Lyon, co-supervised with Quentin Bertrand, since April 2026.
• Tom de Oliveira, M2 intern from Sorbonne Université, co-supervised with Quentin Bertrand and Rémi Emonet, since April 2026.

Alumni:

• Can Pouliquen, PhD student from Polytech Montpellier cosupervised with Titouan Vayer and Paulo Gonçalves, Nov. 2022 - Dec. 2025.
• Maël Chaumette, M2 intern from ENSAI co-supervised with Rémi Gribonval, April 2024 - Sept. 2024, now PhD student in the Ockham team.
• Florian Kozikowski, M1 Intern from Polytechnique, April 2025 - Aug. 2025.
• Ilias Bouhss, L3 Intern from ENS Paris-Saclay co-supervised with Ségolène Martin and Anne Gagneux, June 2025 - July 2025.
• Wassim Mazouz, M1 Intern from Centrale Lyon co-supervised with N. Pustelnik, May 2024 – Aug. 2024.
• Badr Moufad, research engineer, April 2022 - Dec. 2023. Now Phd student at Ecole Polytechnique.
• Pierre-Antoine Bannier, Master thesis for Polytechnique/HEC, May 2022. Now Research scientist at Owkin.

News

• 07/26: I will give a tutorial on memorization and generalization of flow matching and diffusion at ICML 2026 in Seoul, with Quentin Bertrand
• 06/26: Happy to have been a jury member for Inria’s ISFP positions, on benchmark and safety of AI.
• 06/26: With Kimia Nadjahi, Elsa Cazelles and Hadrien Hendrikx we organize the 4th edition of Learning and Optimization in Luminy (LOL) in CIRM (Marseille). This edition’s themes are Generative modelling, optimization for deep learning, and trustworthy ML.
• 06/26: In June 2026, with Quentin Bertrand we will organize the Peyresq summer school on generative modelling and optimal transport (in French).
• 02/26: In Feb. 2026, with Audrey Repetti and Carola-Bibiane Schonlieb, we organized a one week AI hackathon for women in mathematical sciences, during which I gave a 3 h tutorial on generative models (material here).
• 12/25: With Rémi Gribonval we organized a one day workshop on Optimal transport for the departure of Titouan Vayer from the Ockham Team. Best of luck in Rennes!
• 12/25: Our paper on the generalization of flow matching was presented as an oral at Neurips in San Diego (top 0.3% submitted papers)
• 10/25: I have organized a 1 day workshop on generative modelling, diffusion and flow matching on October the 24th at ENS de Lyon.
• 06/25: I am now the Secretary of the MODE groupe of SMAI (which is the Optimization group of the French Society for Applied and Industrial Maths).
• 06/25: With Quentin Bertrand we organized a minisymposium on generative modelling, optimal transport and image restauration at SMAI 2025, June 2nd-6th.
• 03/25: We organized a thematic day on bilevel optimization at ENS Lyon with Jordan Frécon and Quentin Bertrand
• 02/25: With Anne Gagneux, Ségolène Martin, Rémi Emonet and Quentin Bertrand, our blog post on Flow matching was accepted at ICLR 2025’s blog post track.
• 06/24: I am happy to have joined the committee of SMAI’s MODE group (optimization-related group of the French Society for Applied and Industrial Mathematics).
• 06/24: With Elsa Cazelles, Thomas Moreau, Aymeric Dieuleveut and Lorenzo Rosasco we organized LOL 2024 (Learning and optimization in Luminy) in CIRM. It was a blast, we will hold the 4th edition in June 2026: stay tuned!
• 03/24: We organized the SMAI MODE days in Lyon, from march 27th to march 29th. A minicourse on optimal transport by Gabriel Peyré and Yann Brenier was held at ENS de Lyon on the 25th and 26th.
• 10/23: With Titouan Vayer and the help of GDR MIA we organized a thematic day on dimension reduction at ENS Lyon.
• I gave a course on convex optimization at the Computation and Modelling school in Wrocław; resources are on my teaching page.
• celer 0.7 is released, with a fast ElasticNet solver thanks to Badr Moufad!
• We have integrated skglm into scikit-learn, providing a customizable and accelerated solver for sparse GLMs in python.