AI Research Scientist

Meta is recruiting a Research Scientist to join its Fundamental AI Research (FAIR) team in Paris. This full-time, onsite role centers on advancing the reasoning abilities of large language models, with a focus on reinforcement learning, both informal and formal mathematics, and scaffolding or agentic techniques. You would become part of a broad, interdisciplinary group of scientists and engineers, with access to the company’s resources and research facilities.

The position is a fit for someone who enjoys tackling hard theoretical and applied problems, collaborating across disciplines, and turning fresh ideas into concrete experiments, meaningful results, and shared code. If you value rigorous work, mentorship, and cross-functional teamwork, this is a place where your efforts can influence both research directions and potential product implications.

Based in Paris, you’ll operate in a real-world research setting that emphasises collaboration with engineers, product partners, and other researchers, all while having the space to pursue ambitious questions within LLM reasoning.

Reasoning at scale in Paris

In this role, you’ll lead and collaborate on research that pushes forward how language models reason, leveraging reinforcement learning and mathematical reasoning to build better, more capable systems. You’ll set long-range aims and map out intermediate milestones to keep projects focused and measurable.

Your day-to-day will be hands-on: you’ll help design experimental details, implement reusable code, run thorough evaluations, and organize results so findings can be shared across teams. You’ll contribute to publications and open-source projects, while mentoring colleagues and collaborating with cross-functional partners to move ideas from concept to impact.

Must-have credentials

To be considered for this role, Meta is looking for candidates who meet these criteria.

• You hold a PhD in computer science, mathematics, or a closely related quantitative field, or you are in the process of earning one
• You have first-author publications at peer-reviewed AI conferences such as NeurIPS, ICML, or ICLR
• You bring practical experience training, fine-tuning, or experimenting with foundation models beyond using them as black-box tools
• You have worked with state-of-the-art reinforcement learning codebases and are familiar with at least one deep learning framework (for example PyTorch or VERL)
• You must have work authorization for France at the start of employment and maintain authorization throughout your tenure
• You are familiar with LEAN principles

Compensation and practical details

The role is full-time and onsite in Paris, France. A valid work authorization for the country is required from the start date and must remain valid during employment.

How to stand out for this FAIR role

Lead with your strongest credentials by placing your PhD focus and top-tier publications at the top of your résumé, so reviewers see your alignment with this research track right away.

Demonstrate concrete RL and foundation-model experience in your project history, including details about the models you trained or refined, the frameworks you used, and the results you achieved, with clear metrics.

Be ready to walk through how you design experiments, implement reusable code, and document results so others can reproduce and build on your work. Show how you’ve organized data, experiments, and evaluations to support reproducibility and collaboration.

Think of a smart question to ask during interviews that shows your interest in FAIR’s current focus and how you would collaborate with cross-functional teams on long-term goals and milestones.