RESEARCHED WITH AUTUMN
Joachim Studnia
AI Scientist at Mistral AI | Stanford MS '24
Mistral AI · San Francisco Bay Area · engineer
France IMO 2017 Team MemberConcours General 2017 Physics-chemistry PrizeCo-author, Stanford CS224N Multi-Task BERT Report (2023)Co-author, Mistral AI Model Papers (Pixtral 12B, Magistral, Ministral 3)
French AI researcher and engineer; IMO 2017 France team member; Concours general physics-chemistry laureate (2017); Stanford ICME graduate; AI Scientist at Mistral AI.
Machine learningNatural language processingMultimodal / vision-language modelsMathematics
- Quant employment verification - UNRESOLVED: (1) Qube Research & Technologies is a genuine quantitative/systematic investment manager ('a global multi strategy investment manager', qube-rt.com) - so the employer classification fits. (2) QRT itself posted a LinkedIn congratulation about Joachim Studnia, indicating a real QRT association, though the feed text was not retrievable. (3) The specific 'Quantitative Research Intern, Apr-Aug 2022' title/dates come only from the caller's unverified lead. (4) No other quant-finance employer (bank/asset manager/energy/crypto trading) was found. (5) No single source ties the 2017 Olympiad teenager to the current Mistral AI scientist.2022
- Olympiad identity: named to France's six-person team for the 2017 International Mathematical Olympiad (Rio de Janeiro, 13-23 July 2017) by Animath, the French maths-olympiad organiser.Jul 2017
- Co-authored Mistral AI model papers: Pixtral 12B (arXiv 2410.07073, Oct 2024), Magistral (arXiv 2506.10910, Jun 2025) and Ministral 3 (arXiv 2601.08584, Jan 2026).Oct 2024
- Pursued graduate study at Stanford (ICME), where he co-authored a Stanford CS224N (Winter 2023) final report on Multi-Task BERT; profile of 'Stanford MS '24' on LinkedIn.2023
- Qube Research & Technologies describes itself as 'a global investment manager where we deploy a diverse range of investment strategies across geographies, asset classes and time frames' - confirming it operates in quantitative/systematic investment management.
- Co-authored 'Around the BERT model' - a Stanford CS224N (Winter 2023) final report on Multi-Task BERT - with Ines Dormoy and Yoni Gozlan, all listed under Stanford ICME.2023
- Won a prize in the Concours general (physics-chemistry) in 2017 as a Terminale scientifique student at Lycee Condorcet, Paris; L'Express reported he was 16 and had skipped a grade.Jul 6, 2017
- ResearchGate auto-generated author page lists 6 publications attributed to Joachim Studnia (incl. Voxtral Codec work); this is an aggregator, not primary-source confirmation of authorship.
- Ouest-France (18 Jul 2017) published a team photo naming the six-man France IMO 2017 squad: Olivier Garconnet, Joachim Studnia, Timothee Rocquet, Yakob Kahane, Ilyas Lebleu, Baptiste Serraille.Jul 18, 2017
Experience
AI Scientist at Mistral AI
Quantitative Research Intern at Qube Research & Technologies
Apr 2022 - Aug 2022
Graduate student / researcher
Jan 2024
Education
Stanford University
MS
Lycee Condorcet, Paris
Projects
Multi-Task BERT (Stanford CS224N final report, 2023)
Mistral AI model research (Pixtral 12B, Magistral, Ministral 3)
AchievementsSelected for France's national team, 2017 International Mathematical Olympiad · Concours general 2017 prize (physics-chemistry), Lycee Condorcet · Co-author, Stanford CS224N Multi-Task BERT report (2023) · Co-author on Mistral AI model publications (Pixtral 12B, Magistral, Ministral 3)
Articles
AwardsFrance team member, 2017 International Mathematical Olympiad (Rio de Janeiro) · Prize, Concours general 2017, physics-chemistry (Lycee Condorcet, Paris)
Profiles
Competition record
France · IMO