Sayash Kapoor

Sayash Kapoor studies how to evaluate AI agents and frontier systems, building large-scale systems to test their performance and reliability. His research examines gaps between benchmark results and real-world outcomes, including evaluations of AI in scientific research and the reliability of agents.

Career

Cosigned byLoading...

CosignedSign in to view

Elsewhere

Princeton University

Ph.D. in computer science

Products & media

citp.princeton.edu

The Missing Science of AI Evaluation — Princeton final public oral

Kapoor presents research on failures in AI evaluation, including scientific research, agent benchmarks, and methods for assessing real-world performance.

Media appearance