Autocomp is a portable, extensible framework for LLM-driven kernel optimization across tensor accelerators. Point it at a kernel, pick your hardware target, and Autocomp speeds it up, automatically.

Abstract: Since the introduction of Modern Portfolio Theory (MPT) in 1952, its practical applications, associated challenges, and computational efficiency on large and high-frequency datasets, ...

Abstract: Every word or string of words a user types into a search engine has meaning. For example, a user might search for a “hotel” or a “hotel in New York City.” Keywords are the standard focus of ...

I built this project to understand how Markowitz mean-variance optimization actually works in practice - not just the theory, but what happens when you apply it to real assets with correlated returns.