PeopleofSuperintelligence

Researcher

Maxime Darrin

Maxime Darrin is a research scientist at Mistral AI and a co-author of Magistral, the company's first reasoning model, presented in June 2025. Trained in computer science at ENS de Lyon, he added degrees in applied mathematics for machine learning, philosophy of science and public affairs, and interned at Facebook AI in 2021. He began a PhD at Université Paris-Saclay in 2021, joint with Mila at McGill University and supervised by Pablo Piantanida and Jackie CK Cheung, and worked on out-of-distribution detection with RainProof and on summarization with GLIMPSE and COSMIC. His later work includes the Voxtral audio models, the Ministral 3 model family and, in 2026, the Robostral Navigate robot-navigation model and the Shieldstral safety classifier.

Works
30
Citations
58
h-index
4
Badges

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2017–2021

    Computer science at ENS de Lyon

    He studies computer science at ENS de Lyon from 2017 to 2021 and earns a master's degree in fundamental computer science. Alongside it, he completes a master's in applied mathematics for machine learning at Sorbonne Université, a master's in philosophy of science at Paris 1 Panthéon-Sorbonne, and an engineering degree in public affairs and innovation at Mines Paris.

  2. 2018–2021

    Internships and Facebook AI

    Research internships take him to the CNRS in 2018, Worldline in 2019 and CentraleSupélec in 2020. Between May 2021 and September 2021, he is a research intern at Facebook AI.

  3. 2021–2023

    A joint PhD with Mila

    In December 2021, he starts a PhD at Université Paris-Saclay, and in March 2023 he becomes a PhD student at Mila, in a joint program linking the ILLS, Paris-Saclay and McGill University. Pablo Piantanida and Jackie CK Cheung supervise his research on information theory and machine learning applied to natural language processing.

  4. 2023–2024

    RainProof and OOD detection

    His paper RainProof: An Umbrella to Shield Text Generator from Out-Of-Distribution Data is accepted at EMNLP 2023. AAAI 2024 accepts a second paper, on unsupervised layer-wise score aggregation for textual out-of-distribution detection.

  5. 2024

    GLIMPSE and COSMIC at ACL

    In 2024, ACL accepts two of his summarization papers: COSMIC, on mutual information for task-agnostic summarization evaluation, and GLIMPSE, written with Ines Arous, Pablo Piantanida and Jackie Cheung. GLIMPSE summarizes scholarly peer reviews, using uniqueness scores based on the Rational Speech Act framework to identify relevant sentences.

  6. June 2025

    Magistral at Mistral AI

    In June 2025, he is among the authors of Magistral, Mistral's first reasoning model, built with the company's own scalable reinforcement learning pipeline. The paper presents Magistral Medium, trained on top of Mistral Medium 3 with RL alone, and the open-source Magistral Small.

  7. July 2025–January 2026

    Voxtral and Ministral 3

    In July 2025, he co-authors Voxtral, which presents Voxtral Mini and Voxtral Small, two multimodal audio chat models that understand both spoken audio and text. The Ministral 3 paper of January 2026, which also lists him as an author, introduces dense language models in 3B, 8B and 14B sizes built through Cascade Distillation.

  8. Since 2025

    Research scientist at Mistral AI

    He joins Mistral AI as a research scientist in 2025. In July 2026, he is among the authors of Robostral Navigate, an 8B vision-language model that guides robots from monocular RGB images, and Shieldstral, a 3B-parameter multimodal safety classifier.

Career

  1. Since 2025Mistral AIResearch Scientist
  2. 2023–2024École Normale Supérieure de Lyon
  3. 2021MetaResearch Intern
  4. 2020Université Paris-SaclayResearch Intern
  5. 2019WorldlineResearch Intern
  6. 2018CNRSResearch Intern

Education

  1. From 2023MilaPhD student
  2. From 2023École de Technologie SupérieurePhD student
  3. From 2023McGill UniversityPhD student
  4. 2021École Normale Supérieure de LyonMaster's degree in fundamental computer science
  5. 2021Sorbonne UniversityMaster's in applied mathematics for machine learning
  6. 2021Université Paris 1 Panthéon-SorbonneMaster's in philosophy of science
  7. 2021PSL UniversityEngineering degree in public affairs and innovation
  8. From 2021Université Paris-SaclayPhD

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. Classification of red cell dynamics with convolutional and recurrent neural networks: a sickle cell disease case study2023 · Scientific Reports · 26 citations
  2. Unsupervised Layer-Wise Score Aggregation for Textual OOD Detection2024 · Proceedings of the AAAI Conference on Artificial Intelligence · 14 citations
  3. RainProof: An Umbrella to Shield Text Generator from Out-Of-Distribution Data2023 · Conference on Empirical Methods in Natural Language Processing (EMNLP) · 9 citations
  4. Ministral 32026 · 4 citationsWith Abhinav Rastogi, Alexander H. Liu, Alexandre Sablayrolles, Amélie Héliou, Andy Ehrenberg and Andy Lo
  5. Magistral2025 · 3 citationsWith Alexander H. Liu, Alexandre Sablayrolles, Amélie Héliou, Andy Ehrenberg, Baptiste Bout and Baudouin De Monicault

From OpenAlex, as of 21 October 2026.

  1. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Maxime Darrin. The median h-index is 48 at universities and 14 at frontier AI labs. We compare publishing careers, recent citations, dual appointments and job moves across 2,420 university researchers and 308 frontier-lab researchers.
  2. 16% of AI career moves in the 2020s went from a university to a companyIncludes Maxime Darrin. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  3. Companies supply the most prior-work hires at 7 of the 9 frontier AI labsIncludes Mistral AI. Our estimates of the organizations where 550 people worked directly before joining 9 frontier AI labs across 620 career transitions, comparing the main feeder organizations per lab.
  4. Of 943 people shaping AI who worked at Google, 446 work there nowIncludes Mistral AI. Our estimates of where the 943 people shaping AI who worked at Google and Google DeepMind work now: 446 at Google and 452 across 278 other organizations.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Maxime Darrin yet. Send us a free-licensed photo.

  1. Maxime Darrin icannos.github.io. Checked 20 October 2026.
  2. CV | Maxime Darrin icannos.github.io. Checked 20 October 2026.
  3. Maxime Darrin | Mila mila.quebec. Checked 20 October 2026.
  4. GLIMPSE: Pragmatically Informative Multi-Document Summarization for Scholarly Reviews - ACL Anthology aclanthology.org. Checked 20 October 2026.
  5. [2506.10910] Magistral arxiv.org. Checked 20 October 2026.
  6. [2507.13264] Voxtral arxiv.org. Checked 20 October 2026.
  7. [2601.08584] Ministral 3 arxiv.org. Checked 20 October 2026.
  8. Maxime Darrin | alphaXiv alphaxiv.org. Checked 20 October 2026.
  9. [2607.20785] Robostral Navigate arxiv.org. Checked 20 October 2026.
  10. [2607.25857] Shieldstral arxiv.org. Checked 20 October 2026.