PeopleofSuperintelligence

Lab leader

Olivier Duchenne

Olivier Duchenne is a computer vision researcher and AI leader whose work spans graph matching, language models and robotics. He completed a PhD under Jean Ponce at École Normale Supérieure, did a postdoc at Carnegie Mellon with Alexei Efros, and his graph-matching work earned a CVPR honor in 2009. At Meta / FAIR, he contributed to Llama 3 and SWE-RL, then co-authored Magistral and Robostral Navigate at Mistral AI. As of October 2026, he is CEO and co-founder of ORA Robotics, his third startup, which provides robots-as-a-service for manufacturing and aims to build a vision-language-action model.

Now at
ORA Robotics and FAIR (Meta)
Works
32
Badges
Links
Website, Google Scholar and X

Publication numbers from OpenAlex, as of 22 October 2026. Citations and h-index are hidden because this OpenAlex record may be incomplete or mix in other authors.

Timeline

  1. 2008–2012

    PhD at ENS under Ponce

    He earns an M.S. in computer science and applied mathematics at École Normale Supérieure in Paris in 2008, then joins WILLOW as a PhD candidate under Jean Ponce. He defends a 2012 thesis at ENS on non-rigid alignment for object recognition.

  2. 2009–2011

    Graph matching at CVPR

    His tensor-based high-order graph-matching algorithm, co-authored with Francis Bach, In-So Kweon and Jean Ponce, earns a Best Student Paper honorable mention at CVPR in 2009. The work appears in PAMI in 2011; his graph-matching kernel with Armand Joulin and Jean Ponce is an oral ICCV paper in 2011.

  3. 2012

    Postdoc at Carnegie Mellon

    In 2012, he joins Carnegie Mellon as a postdoc in computer vision with Alexei Efros.

  4. 2024

    Llama 3 at Meta

    At Meta / FAIR, he is a core contributor to Llama 3, working on code generation and synthetic data. The paper appears on arXiv in July 2024 and describes a language-model family whose largest dense Transformer has 405B parameters.

  5. 2025

    SWE-RL for software engineering

    In February 2025, he is the second author of SWE-RL, an approach that scales reinforcement learning to real-world software engineering. Llama3-SWE-RL-70B reaches a 41.0% solve rate on SWE-bench Verified, and the work is accepted to NeurIPS 2025.

  6. June 2025

    Magistral reasoning model

    Mistral posts Magistral to arXiv in June 2025 as its first reasoning model, with Duchenne among its authors. The paper presents a scalable reinforcement-learning pipeline built on Mistral's own models and infrastructure.

  7. July 2026

    Robostral Navigate

    In July 2026, Mistral AI introduces Robostral Navigate, its first model built for embodied navigation, and lists him among its authors. The 8B model uses a single RGB camera and achieves 76.6% success on unseen R2R-CE benchmarks.

  8. As of October 2026

    CEO of ORA Robotics

    Mistral AI launches its robotics unit in December 2025; before ORA, he directs robotics there and recruits 15 researchers. In September 2026, ORA's associates name him president; as of October 2026, he is CEO and co-founder. ORA is his third startup after Adriel and Solidware; it provides turnkey robots-as-a-service for manufacturing and aims to build a frontier vision-language-action model.

Career

  1. NowORA RoboticsCEO, Co-founder
  2. 2026Mistral AIDirector of Robotics
  3. From 2024FAIR (Meta)Core contributor
  4. From 2012Carnegie Mellon UniversityPostdoc
  5. AdrielFounder
  6. SolidwareFounder

Education

  1. 2012WILLOWPhD candidate
  2. 2008École Normale SupérieureM.S.

From the biography.

  1. Companies supply the most prior-work hires at 7 of the 9 frontier AI labsIncludes Olivier Duchenne. 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.
  2. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes Olivier Duchenne. We estimate that alumni of 17 major AI labs went on to found or lead 350 other organizations, including 226 they founded, and we trace 472 career moves directly between 23 labs since 2010.
  3. 16% of AI career moves in the 2020s went from a university to a companyIncludes Olivier Duchenne. 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.
  4. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Meta. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

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

  1. A Graph-Matching Kernel for Object Categorization ri.cmu.edu. Checked 21 October 2026.
  2. WILLOW - 2012 - Rapport annuel d'activité radar.inria.fr. Checked 21 October 2026.
  3. WILLOW - 2011 - Rapport annuel d'activité radar.inria.fr. Checked 21 October 2026.
  4. WILLOW - 2012 - Rapport annuel d'activité radar.inria.fr. Checked 21 October 2026.
  5. Alyosha Efros homepage cs.cmu.edu. Checked 21 October 2026.
  6. Olivier Duchenne duchenne.net. Checked 21 October 2026.
  7. The Llama 3 Herd of Models arxiv.org. Checked 21 October 2026.
  8. SWE-RL: Advancing LLM Reasoning via Reinforcement Learning on Open Software Evolution arxiv.org. Checked 21 October 2026.
  9. Magistral arxiv.org. Checked 21 October 2026.
  10. Robostral Navigate: single-camera AI navigation | Mistral AI mistral.ai. Checked 21 October 2026.
  11. Robostral Navigate arxiv.org. Checked 21 October 2026.
  12. ORA Robotics (Le Figaro Entreprises) entreprises.lefigaro.fr. Checked 21 October 2026.
  13. Ora — Companies — The Long Build thelongbuild.eu. Checked 21 October 2026.