Language is infrastructure. I study how machines learn to read it — and how economies are rewritten by the ones that do.
Student researcher in NLP, computational economics and human-centered AI. Co-founder & CFO of LinkScape. Shanghai, China.
THE STRUCTURE BEHIND CNM-BERT
Two papers, readable here.
Each is published twice: a plain-English page under its section, and the manuscript itself with the PDF alongside. Both, side by side →
A Chinese character is built from parts, and a tokenizer throws every one of them away. This paper hands the parts back to the model, and watches what happens to the characters it has never seen.
One hundred million job adverts, read one at a time by a language model, then turned into a firm-by-firm record of who is adopting AI — and what it does to the work they hire for.
Five sections, and a way to reach me.
CNM-BERT, and why a model that can see inside a character stops guessing when it meets a rare one. Plus PoliScope, a RAG system over US legislative and campaign-finance data.
01 · LANGUAGE →The AIFE Engine — measuring AI adoption from hiring language across 72 countries, then asking whether it causes the shift away from routine work. Olympiad medals sit here too.
02 · ECONOMY →LinkScape, the student nonprofit I help run, and Resonaite, its adaptive-audio flagship. Also VitaTex, Ring-Ing, and the hackathons in between.
03 · BUILDING →Three Scholastic Gold Keys, a quarterfinal at the Harvard World Schools Debate Open, and a standing habit of reading well outside my field.
04 · WORDS →School, leadership, the toolbox, and what I do when none of the above is open — rowing, historical fencing, and a great deal of LEGO.
05 · PERSON →Researchers, founders, admissions readers — the door at the bottom of every page is open.
06 · CONTACT ↓