AI Research Scientist, Multi-Agent Memory - FAIR, MSL

Meta's Fundamental AI Research (FAIR) organization is seeking a Research Engineer to join the team working on multi-agent systems and memory. The role involves driving fundamental breakthroughs to enable agents to reason, learn from experience, and collaborate on complex tasks, contributing across the full spectrum of research, engineering, and deployment for both product and frontier model efforts.

Responsibilities

• Design and execute large-scale research experiments to push the state-of-the-art in machine intelligence, language models, multi-agent communication, test-time scaling, and recursive self-improvement
• Steer projects from early exploration through evaluation on benchmark datasets such as SWE-bench, RSI-Exam, and PaperBench
• Directly contribute to experimental design, harness engineering, maintaining large codebases, running evaluations, and tracking end-to-end performance
• Publish novel algorithmic techniques at top-tier venues, foster open joint experimentation, and mentor team members
• Lead cross-functional research initiatives and establish solution recipes that bridge research clusters to production serving environments

Qualifications

• Bachelor's degree in Computer Science, Computer Engineering, relevant technical field, or equivalent practical experience
• Bachelor's degree in Computer Science, Computer Engineering, a relevant technical field, or equivalent practical experience
• A PhD degree in Computer Science, Mathematics, or a similar discipline
• Proven track record of publications at top-tier AI and NLP conferences (e.g., NeurIPS, ICML, ICLR, ACL, NAACL, EMNLP)
• Demonstrated experience with training, fine-tuning, and experimentation on language models and Transformers beyond black-box usage
• Must be able to obtain work authorization in the country of employment Experience working with deep learning frameworks, building research prototypes, and communicating cross-functionally in a team environment
• Research experience in long-term memory, multi-agent systems, agent orchestration, self-improvement, or automated scientific research
• Experience with large language model pre-training, fine-tuning, long-context modeling, context compression, or evaluation pipelines at scale
• Experience building, evaluating, and designing controlled experiments for language-model agents, including tool use, memory mechanisms, or collaborative workflows
• Proven track record of achieving significant research results, as demonstrated by publications at leading AI and NLP conferences