10 Cohere research scientists
Half of these Cohere research scientists describe post-training work, from RLVR and reinforcement learning to a post-training recipe for agentic AI, and three hold lead or head titles at Cohere today. They suit a team choosing how to post-train or evaluate a model for enterprise use, or weighing multilingual and reinforcement learning methods, and between them they have also worked at Google DeepMind, AWS, Ai2 and NAVER LABS Europe.
Professionals to explore
Alexandre Bérard
LinkedInExperience: Research Scientist · Cohere
$1,010 per 30-minute call
Alexandre Bérard has been a Research Scientist at Cohere since 2024, working on Multilinguality. NLP Research Scientist at NAVER LABS Europe (2018-2024) and PhD Student roles at the Laboratoire d'Informatique de Grenoble (2016-2018) and Inria (2014-2016) make Alexandre the multilingual specialist here, the one to ask about language coverage in an enterprise model.
Is this you? Claim your profile and get paid for calls.
Beyza Ermiş
LinkedInExperience: Senior Research Scientist · Cohere
$1,346 per 30-minute call
Beyza Ermiş has been Senior Research Scientist at Cohere since 2023 and was a Research Scientist at Amazon Web Services (AWS) from 2017 to 2022, with a Research Assistant role at Boğaziçi University (2012-2017) and two Amazon machine learning internships. Research scientist titles at two companies suit questions comparing research scientist work at AWS and at Cohere.
Is this you? Claim your profile and get paid for calls.
Florian Strub
LinkedInExperience: Senior Research Scientist · Cohere
$1,346 per 30-minute call
Florian Strub describes co-leading Cohere's Post-Training department and leading a team of 15 people in charge of RLVR training and the engineering of the post-training stack, as Head of RLVR and Post-training enginering at Cohere since 2025. Co-head of Command A and Command R7B Post-training (2024-2025) and Senior Research Scientist at Cohere (2024) and at Google DeepMind (2019-2024) make Florian a fit for questions on post-training the Command models.
Is this you? Claim your profile and get paid for calls.
Marzieh Fadaee
LinkedInExperience: Staff Research Scientist · Cohere
$1,683 per 30-minute call
Marzieh Fadaee has been Head of Cohere Labs since September 2025, following Staff Research Scientist (2025) and Senior Research Scientist (2023-2024) at Cohere. NLP Research Lead (2021-2023) and NLP ML Research Engineer (2019-2021) at Zeta Alpha complete a record that fits questions on organizing a research group such as Cohere Labs.
Is this you? Claim your profile and get paid for calls.
Maximilian Mozes
LinkedInExperience: Post-Training Team Lead · Cohere
$1,346 per 30-minute call
Research Scientist (2023-2024) and Senior Research Scientist (2024-2025) at Cohere, and Post-Training Team Lead there since April 2025, Maximilian Mozes is the person to ask how a post-training team is organized. Their LinkedIn profile gives the title Staff Research Scientist at Cohere, and before Cohere, Maximilian was Postdoctoral Researcher at Tilburg University and Student Researcher at Google (both 2022-2023).
Is this you? Claim your profile and get paid for calls.
Nouha Dziri
LinkedInExperience: Senior Research Scientist · Cohere
$1,346 per 30-minute call
Nouha Dziri has been Senior Research Scientist at Cohere since March 2026, the newest Cohere start on the page. Research Scientist (2024-2025) and Postdoc (2022-2023) at Ai2, Visiting Researcher at Mila - Quebec Artificial Intelligence Institute (2021-2022) and AI Researcher at Google (2019-2022) give Nouha a view of research at Ai2, Mila and Google as well as Cohere, useful for comparing research settings.
Is this you? Claim your profile and get paid for calls.
Pat Verga
LinkedInExperience: Former Senior Research Scientist · Cohere
$1,346 per 30-minute call
Pat Verga was Senior Research Scientist at Cohere (2023-2024) and has also held Staff Member of Technical Staff (2024-2025) and Senior Staff Member of Technical Staff (since 2025) there. Other titles include Senior Research Scientist at Google DeepMind (2023) and Research Scientist at Google (2019-2023), and Pat describes more than 10 years of experience in NLP and Machine Learning, a fit for questions on NLP research practice at Google and Cohere.
Is this you? Claim your profile and get paid for calls.
Pierre Clavier
LinkedInExperience: Member of Technical Staff · Cohere
$808 per 30-minute call
Pierre Clavier is a Member of Technical Staff at Cohere (since 2025), following a Research Scientist Intern role there (2024-2025), and describes the work as large language model post-training and reinforcement learning research. A PhD at Inria (PHD Candidate, 2021-2024) that focused on the theory and practice of robust reinforcement learning makes Pierre a fit for questions on robust reinforcement learning methods in post-training.
Is this you? Claim your profile and get paid for calls.
Yannis Flet-Berliac
LinkedInExperience: Research Scientist · Cohere
$606 per 30-minute call
As Research Scientist at Cohere since 2024, Yannis Flet-Berliac focuses on large language model post-training and reinforcement learning research. Research Scientist at InstaDeep (2023-2024), a Stanford University postdoc (2022-2023), Ph.D. Candidate at Inria (2018-2021) and an ICML Associate Program Chair role point to questions on applying reinforcement learning in post-training.
Is this you? Claim your profile and get paid for calls.
Yixuan Su
LinkedInExperience: Staff Research Scientist · Cohere
$1,683 per 30-minute call
Agentic reasoning is Yixuan Su's area: Staff Research Scientist at Cohere since July 2026, after Senior Research Scientist (2025-2026) and Research Scientist (2023-2025) there, and describes leading the modelling efforts on agentic reasoning and developing a post-training recipe for agentic AI. Research internships at AWS AI (2021) and Apple AI/ML (2020-2021) also appear; Yixuan fits questions on training models for agent use cases.
Is this you? Claim your profile and get paid for calls.
Choose the right perspective
For post-training questions, Florian Strub and Maximilian Mozes hold post-training lead titles, Yannis Flet-Berliac and Pierre Clavier describe the reinforcement learning side, and Yixuan Su covers agentic reasoning. Alexandre Bérard is the pick for multilingual work and Marzieh Fadaee for how Cohere Labs is organized, while Nouha Dziri, Beyza Ermiş or Pat Verga can compare research at Cohere with Ai2, AWS or Google.
Questions to take into the conversation
- 01How does a Cohere post-training team decide between RLVR and other fine-tuning methods for a Command model release?
- 02What changes in a post-training recipe when the target is agentic tool use rather than plain chat?
- 03How do Cohere researchers test whether a multilingual model holds up in languages with little training data?
About this directory
This is a professional research starting point based on business profile data retrieved on . Titles and companies reflect that source snapshot; roles marked former had ended by then. Check the linked profiles for current details. Inclusion does not imply Instant Expert membership or availability.