Peter Frazier

May 2026 — Talks in NYC on predicting protein function from evolutionary data at three events: the Frontiers of AI Symposium, the AI in Basic Research, Medicine, and Life Sciences Symposium, and the CECAM Workshop on Physics-Aware Machine Learning. [Paper | Slides]
April 2026 — Inaugural Jeff and Karen Camm Lecture, Clemson University. Talk: Bayesian Preference Exploration: Making Optimization Accessible to Non-Experts. [Slides]
March 2026 — Keynote at the NSF-AIRFoundry Annual Meeting, University of Pennsylvania. Talk: From Nature's Incomplete Dataset to Reproducible Experiments: Addressing Three Challenges in Biomolecular Design [Slides Part 1 | Slides Part 2]
March 2026 — Invited talk at Meta’s Adaptive Experimentation Workshop, New York City. Talk: LLMs as Preference Oracles: Iterative Multi-Objective Selection from Natural Language. [Slides]
February 2026 — Invited talk in the Yale School of Management Operations Seminar Series. Talk: Bayesian Preference Exploration: Scalable Stakeholder Management for Optimizers. [Slides]
January 2026 — Test of Time Award from the INFORMS Journal on Computing for The knowledge-gradient policy for correlated normal beliefs (with W. Powell and S. Dayanik, INFORMS J. Comput. 2009), recognizing foundational work on the knowledge-gradient acquisition function for Bayesian optimization.

My research interests include:

The impact of my work includes:

Better pricing at Uber, where I worked as a scientist from 2015 to 2024
Helping Cornell stay open during the pandemic via a large-scale COVID-19 testing program [PNAS, WSJ, GMA, ABC News] (2020–23)
Logistics decision support software for the US Navy and Marines (2022–now)
Making medical tests and wetlab experiments more reproducible through AI with Saddlepoint Labs (2024–now)