hi, i'm abhishek

i work on mechanistic interpretability and think about the economics of AI. most of my time goes toward understanding what's actually happening inside neural networks. the rest goes toward understanding what happens to markets and institutions when AI reshapes them.

i'm interested in prediction markets, crypto as financial infrastructure, effective altruism, and the places where AI safety and economic policy overlap.

this site is a mix of longer writing and working notes. nothing here is finished. most of it is wrong. but writing in public is how i think.