Senior Applied Scientist / AWS
Agentic RL, LLM post-training, and structured-data foundation models.
About
Physicist by training. Builder by practice.
I work at the boundary between new machine-learning methods, production-scale systems, and scientific questions.
I build learning systems that connect reasoning, data, and the structure of the real world.
I am a Senior Applied Scientist in AWS AGI Foundations. My current work focuses on agentic reinforcement learning, LLM post-training, and foundation models for structured data, including time series and tables.
Before AWS, I was an AI Researcher at Shell, developing physics-informed and weakly supervised methods for scientific data. I earned my Ph.D. in Physics from Harvard University, where I worked with Xiaowei Zhuang on computational and super-resolution microscopy.
That path from molecular-scale imaging to foundation models continues to shape how I approach research: strong methods should respect the data, expose useful structure, and survive contact with real systems.
Applied Scientist / AWS
Foundation-model pretraining, multimodal reasoning, geospatial AI, and model adaptation.
AI Researcher / Shell
Physics-based semi-supervised learning for sparse scientific observations.
Ph.D. in Physics / Harvard University
Machine-learning-driven computational microscopy and quantitative biological discovery.
B.S. in Physics / Shandong University
Foundations in physics, chemistry, and computational science.
Recognition / 01
Awards, talks, and research service.
Area Chair Award
ACL 2024 / CaMML
Invited research talks
ICLR Expo, NeurIPS Expo, ICML Expo, and LBNL AI4Science
Workshop organizer
ICLR AI4Differential Equations and ICML Foundation Models for Structured Data
Research community
Reviewer for ICLR, ICML, NeurIPS, KDD, CVPR, ICCV, and ARR
Distinction in Teaching
Harvard University / 2014
Karplus Prize
Harvard University / 2013