Brian Zhou builds AI-agent evaluations at Thoughtful, including BallotBench’s live election-forecasting benchmark, and co-founded a37 to build coding agents for infrastructure. On leave from Harvard, he also founded SPRING, coaches the U.S. AI Olympiad team and has researched machine learning, robotics, public policy and redistricting at Harvard, MIT, Brookhaven, Dartmouth and the Naval Research Laboratory.
BallotBench Creator
Built a live evaluation of frontier AI agents forecasting the 2026 U.S. midterm elections.
Co-founder
Built coding agents for infrastructure management; currently listed as co-founder on his personal work page.
Founder and Chair
Scientific Committee Member and Team USA Coach
Research Partner and Research Scholar
HKS/Radcliffe research partner; IQSS research scholar. Earlier: TECH Program Director (May 2024–May 2025), HKS researcher (Sep 2024–Mar 2025), Radcliffe AI intern (Aug 2024–Mar 2025), statistics intern (Apr 2023–Mar 2025).
Machine Learning Research Intern, CSAIL
Quantum Information Science and Engineering Intern
Machine Learning Research Intern
Student Researcher, Machine Learning and Robotics
Worked at the U.S. Navy’s research laboratory; earlier Deep Learning Research Intern, May–August 2022 and May–August 2023.
Teaching Fellow, Research Statistics III
Vice President of Operations
Computer Science and Statistics
