I work in both theory [statistics/information theory] and engineering [computer systems]. I was affiliated to the Stanford Statistical Machine Learning Group, Stanford Platform Lab, and the Information Systems Laboratory (ISL).
After leaving academia, I became a trader/speculator, mostly expressing my semi-systematic views in commodities, taking flat-price and spread risks across major energy, metals, and agriculture markets. With statistical methods, machine learning, and more recently the new wave of agentic AI, I aim to competitively serve the market and its participants — contributing to better price discovery and more efficient risk transfer.
I work in the areas of machine learning and natural language processing. While I am no longer engaged in frontline large language model research, my earlier work contributed to both academic and industrial applications of language models, including some of the earliest integrations of LLMs into search engines in 2016. I broadly identify with the machine learning research community, particularly COLT, ICLR, ICML, and NeurIPS, and maintain strong interests in statistics and information theory. For a period in the past, I also conducted research in systems and networking.
My research has received recognition in both academia and industry. Data center networking research was featured on the front page of The New York Times, and has led to the creation of multiple startups. In machine learning, my work in natural language processing has been taught in Stanford University's widely attended CS224n course, led by Professor Christopher Manning.
I serve as a reviewer for several major conferences, including KDD, NeurIPS, ICLR, and ICML.