Research Scientist focused on mechanistic interpretability of large language models to ensure safety, alignment, and reliability. This role involves exploring emerging interpretability methods, developing open-source infrastructure, performing causal validation, and publishing findings. The scientist will also write code for experiments on distributed compute clusters.

Applied Research role focused on foundational audio algorithms for Pixel and Buds, specifically Superhuman Hearing and Open Ear ANC. The role involves exploratory, non-timeline-based research in a high-velocity, startup-style team, aiming for massive user differentiation and defining agentic audio experiences.