About

I'm a hybrid product manager and research scientist (and doing a PhD part-time at University of Cambridge [for the fun of it!]) working at the intersection of artificial intelligence, psychiatry, and safety.

As a PM, my focus has been on shipping ML/LLM systems in the mental health space. At Tempus AI, I lead our AI biomarker, psychiatry, and provider genAI product teams. Previously, I was Product Lead for Psychiatry AI/ML at Verily (Google Life Sciences), owning the behavioral-health portfolio across four cross-functional teams — including AI/ML digital biomarkers across linguistics, paralinguistics, and mobile sensors and our work building novel EHR suicidality and self-harm risk detection models.

At the same time, I'm a part-time PhD candidate in Psychiatry at the University of Cambridge, focused on applied AI and computational psychiatry. My research focuses on LLM safety and safeguards for psychiatric use — susceptibility/risk biomarkers for LLM-induced, low-base-rate adverse events, and empirical model-behavior evaluations (including sEEG measurement of sycophancy) that distill the drivers of safety risk in LLM use. Before that I was a graduate researcher at Stanford focusing on interpretable (XAI and FAccT) deep-learning models for PTSD and schizophrenia risk detection.

I hold an M.S. in Computer Science (Biocomputation) and a B.S. in Symbolic Systems (AI) from Stanford University.

Education

Experience

Advisory & Steering Committees

Publications