Recent
Research
A unified model for GUI grounding across web, mobile, and desktop. Trained on large-scale synthetic data. SOTA on all major GUI grounding benchmarks.
Exploring efficient and effective LLM reasoning with reinforcement learning from verifiable rewards. Microsoft Research Asia.
Enhancing e-commerce content moderation with large language models. Production-scale deployment.
Building
Early-stage AI startup. Building in stealth with CMU venture studio.
Natural language workflow automation engine — YC Browser Use Hackathon.
2nd place winner at AdventureX 2025 hackathon.
Personal lifestyle assistant for budget-friendly living. 2nd place at Baidu AGI Hackathon.
MS student at CMU Language Technologies Institute, graduating December 2026. I'm obsessed with making AI agents that actually work on the open web — and quietly plotting my own startup on the side.
Before CMU: RLVR optimization at MSRA, multimodal ML at ByteDance, and grounding research at OSU NLP Group.