Research Engineer - PAR Safety

Meta is seeking Research Engineers to join the Data & Privacy Research team within Meta Superintelligence Labs. This team studies, evaluates and aligns Meta's frontier AI model and systems, with a focus on memorization & privacy risks. We ensure that our frontier models and agents operate and behave within users' privacy and Meta's privacy expectations by rigorously measuring and mitigating across the entire frontier model development lifecycle - pre-training data curation to post-training alignment. As a Research Engineer on this team, you will work alongside AI researchers to develop new evaluations grounded in real-world threat models, maintain existing evaluations so they remain current and reliable, and produce written artifacts that Meta can trust during high-stakes launches. This is a highly technical role requiring the ability to deeply understand frontier LLM behavior, hypothesize and design true and grounded risk vectors and turn them into automated evaluations with high scalability and reliability. The evaluations you build will directly inform risk assessments and launch decisions within Meta Superintelligence Labs, making engineering reliability, rigor, and scalability paramount. You will succeed by delivering iteratively while re-prioritizing work based on evolving research needs and launch timelines as we advance the technical research frontier. The evaluations you produce will significantly influence pre-training data mixes, model behavior during post-training and will be read and acted upon by Meta's leadership during model launches and policy reviews.