Privacy Research Engineer, Safeguards

Lead the privacy analysis of frontier models, auditing data use and ensuring safety throughout the development cycle. Design and implement privacy-first training algorithms and techniques for large-scale models. Develop evaluation and auditing methods to measure the privacy of training algorithms and data handling. Collaborate with a senior team to define forward-looking privacy policies and advocate for responsible data use on behalf of users. Build tooling and processes for privacy-preserving ML, including differential privacy approaches and clear distinctions from k-anonymity, l-diversity, and t-closeness. Bring strong software engineering skills (Python, PyTorch or JAX) and experience shipping in fast-moving startups, with a track record of leading cross-functional security initiatives.