AI Research Scientist - MSL FAIR Alignment

Job description

Meta is seeking AI Researchers to join the FAIR RAM (Reasoning, Alignment, Memory) team, part of the FAIR pillar in Meta Superintelligence Labs. Our team pursues curiosity-driven research to develop learning algorithms with enhanced reasoning, memory, and alignment. Current projects span long-horizon reinforcement learning + agents, improved learning + self-supervised learning objectives, higher-level reasoning, new memory techniques, and scalable alignment methods including self-alignment, self-improvement, and co-improvement.

Our team is a research-centric team working across both pure research and research-to-product (R2P). We conduct cutting-edge fundamental AI research with industry-scale resources, and we openly publish our results.

Responsibilities

Define and drive research agendas in reasoning, memory, alignment, and agentic learning. Design and run large-scale experiments to test novel algorithmic ideas, including long-horizon RL, self-supervised objectives, and tool-using agents. Develop new methods for scalable alignment - self-alignment, self-improvement, and co-improvement loops that reduce dependence on human-annotated supervision. Invent and evaluate new memory architectures and techniques for models that reason over long contexts and extended horizons.

Conduct cutting-edge fundamental AI research, publishing at top-tier venues, and contribute to the research community through reviewing, talks, and collaboration. Partner closely with other fundamental research teams in MSL, and research-to-product teams in other pillars.

Qualifications

• Bachelor's degree in Computer Science, Computer Engineering, relevant technical field, or equivalent practical experience PhD in Computer Science or a related technical field
• 2+ years of industry research experience in Generative AI and LLM 1+ year of experience as a formal technical lead, leading major technical initiatives with cross-functional impact, and/or influencing strategy across multiple teams
• Research experience in LLM post-training and/or agents Published research in leading peer-reviewed conferences (e.g., ACL, EMNLP, NeurIPS, ICML, ICLR) and/or demonstrated significant industry influence in the field of AI Proficiency in Python and a modern deep learning framework (e.g., PyTorch) Domain expertise in LLMs for agents, higher-level reasoning, alignment methodologies, and/or new memory techniques First-author publications at top peer-reviewed conferences (e.g., ACL, EMNLP, NeurIPS, ICML, ICLR) Experience working on frontier-quality, state-of-the-art Large Language Models