Research Scientist, AI Verification

Meta is looking for Research Scientists to join our AI Verification team. For these positions we are looking for people with experience in model training or agentic AI to complement the verification expertise in our team. The candidate will have experience in building trustworthy and reliable AI models and agents, including designing comprehensive evaluation benchmarks, optimizing training processes, and ensuring high-quality data pipelines. The candidate will help to create and evaluate data and environments to train ML models, and benchmarks to judge agents and models. The overall focus of the team is on using formal program verification to make AI better (safer, more reliable), and on using AI to make formal verification better (more feasible); these ML roles will be pivotal in connecting symbolic verification to neural AI.

Responsibilities:

Lead, collaborate, and execute on research that pushes forward the state of the art in Ai Reasoning and Verification Collaborate with verification experts in the team as well as machine learning and agentic experts across AI research teams across Meta Work towards long-term ambitious research goals, while identifying intermediate milestones Directly contribute to experiments, including designing experimental details, implementing reusable code, and designing and running evaluations Contribute to publications and open-sourcing efforts Mentor other team members and collaborate with cross-functional partners to align on research goals and deliverables

Qualifications:

Currently has or is in the process of obtaining a PhD in the field of Computer Science, Artificial Intelligence, Machine Learning, or a related field Experience in ML model evaluation, training pipelines, or benchmark and agent design and evaluation Publications in areas of Machine Learning, AI, or related fields Must obtain work authorization in the country of employment at the time of hire, and maintain ongoing work authorization throughout employment First-author publications at peer-reviewed AI conferences (e.g. NeurIPS, ICML, ICLR, \*ACL, EMNLP) Experience with large-scale training infrastructure and data pipelines Experience designing and implementing evaluation benchmarks for large AI models or agents Experience with formal verification or automatic program analysis