AI Research Engineer @ Meta
I work on AI systems — from ranking and recommendation to LLMs and agents. I like thinking about how models behave in messy, real-world environments. Previously worked at Google and Amazon Alexa AI. Visiting Scholar at Harvard and University of Waterloo.
I write about things I'm learning and researching. Here's what I've been covering on my blog:
🛡️ AI Safety & Alignment — How reward hacking evolved from classical RL specification gaming to jailbreaks and deceptive alignment in LLMs. What it means for RLHF and building systems we can trust.
🧠 LLM Reasoning — What "reasoning" actually means in the context of large language models, grounded in research from chain-of-thought prompting to inference-time compute scaling.
🌐 Browser Agents & Goal Fidelity — Why the web is an adversarial environment for agents, and why being capable is not the same as being hard to manipulate.
🎯 Ranking & Recommendation Systems — A deep-dive series covering the full evolution: from foundational collaborative filtering, through the deep learning era, to modern sequential learning and long user history modeling in ads systems.
When I'm not thinking about AI:
🏂 Snowboarding — PSIA-AASI Level 1 certified instructor