I like turning difficult ideas into things that can be examined, used, and improved. This profile is a record of what I have built across research, systems, experiments, and open-source work.
ml-quant-trading is an end-to-end
PyTorch research stack for machine-learning multi-factor trading. It brings together
213 factor dimensions, bias correction, ML baselines, portfolio optimization,
vectorized backtesting, public-data validation, tests, and benchmark tooling.
Run it in Colab · See validated results · Read the paper · Contribute a reproduction
The project is validation-first: it documents costs, turnover, baselines, data limitations, and negative results rather than presenting backtests as investment claims.
A few other projects that represent different stages and interests.
| Project | A short note |
|---|---|
| me_fasttext | An exploration of memory-efficient text classification and compact data structures. |
| monkey | A small compiler back end, built to understand how language becomes execution. |
| tiger | A small compiler project built to explore language design from source to execution. |
| dbms | A focused experiment in database internals and buffer management. |
Building, testing, and documenting ideas in public.