scroll ▾
I'm a recent graduate of Carnegie Mellon University's MSCS program (2026), where my research focused on Multi-Agent LLM Systems and AI Safety.
My master's thesis examined deceptive behaviors in multi-agent LLM systems, co-advised by Vincent Conitzer (CMU) and Zhijing Jin (University of Toronto). I also work remotely as an external collaborator for the Jinesis Lab.
During my undergrad at CMU, I completed majors in both Computer Science and Mathematics with a minor in Computational Finance, graduating with University and College Honors.
→ Presenting When Agents Lie (Best Paper Award) at the ICML NExT-Game workshop in Seoul on July 11, 2026.
→ Joining The Voleon Group as a Software Engineer in July 2026.
→ Open to summer lecturer positions in CS, ML, AI, or quantitative finance.
→ Passionate about teaching: TA for 7 different CMU courses, from differential equations to graduate AI. See my teaching philosophy.
→ Off-hours: tricking, breakdancing, freediving, photography, and K-Pop dance — proof on the hobbies page.
Best Paper Award. "When Agents Lie: Premeditation, Persistence, and Exploitation in Repeated Games" accepted to the ICML New Frontiers in Game-Theoretic Learning (NExT-Game) workshop — and won the Best Paper Award. Presenting in Seoul on July 11. Read the paper.
Paper accepted. "What Game-Theoretic Benchmarks Miss: Strategic Silence in Multi-Agent LLMs" accepted to the ICML Workshop on Failure Modes of Agentic AI. Read the paper.
Thesis defended. Successfully defended my master's thesis, "The Structure of Deception: How LLM Agents Lie, Break Promises, and Exploit Trust in Multi-Agent Settings." Watch the defense presentation.
Paper accepted. "Cheap Talk, Empty Promise: Frontier LLMs easily break public promises for self-interest" accepted to the ICLR AI for Mechanism Design and Strategic Decision Making workshop.
Paper accepted. "Behavioral and Strategic Deception in Large Language Models: A Taxonomy and Benchmark Analysis" accepted to the ICLR Agents in the Wild: Safety, Security, and Beyond workshop.
Paper accepted. "Market-Dependent Communication in Multi-Agent Alpha Generation" accepted to the NeurIPS GenAI in Finance workshop.
How do LLM agents behave when they can communicate, commit, and privately deviate? My work builds taxonomies, benchmarks, and empirical evaluations of deceptive behavior in multi-agent systems.
Premeditation, persistence, and exploitation: how LLM agents deceive in repeated games. Best Paper at the NExT-Game workshop.
Frontier LLMs break public commitments in ~57% of scenarios — often without recognizing they're doing it.
A unified taxonomy of LLM deception, applied to 35 benchmarks to expose systematic coverage gaps.
Imagining a future classroom where AI tools enhance learning through collaborative homework, conversational exams, and authentic project-based assessment.
2025.09.08A practical walkthrough of using AI to understand dense academic papers, from adversarial training to developing genuine mathematical intuition.
2025.09.01 · featuredExploring how AI's current failures in mathematics reveal the nature of intelligence, and why this window of distinguishability is rapidly closing.
A structured learning platform for aspiring quant professionals — a skill tree from calculus to frontier research, with AI tutoring at every level. Designed to help you land your first quant role without the six-figure MFE commitment.
Join the waitlist