Semantic Movie Search

Semantic Movie Search1 is a search engine for movies which tries to predict what movie you are thinking of based on queries involving vaguely rememebered plot points, e.g. "cowboy doll gets jealous of spaceman toy" will give you Toy Story. I built this search engine along with Anna Brosowsky, Nancy Wang, Ethan Zell, and Yili Zhang for the Erdös Data Science bootcamp, where this project won the best project prize. The source code for the search engine, as well as the datasets, can be found on Github.

Evil Set

Evil Set is a variant of the card games Set and Ultra Set which randomizes the attributes that vary across the deck for each game. The randomized attributes get rid of the muscle memory aspect of the game, and levels the playing field between new and experienced Set players.


Org-to-Things is a filter written in Haskell for Pandoc that turns Org files to links that can be imported into the iOS planner app Things.

Word problem for hyperbolic groups in CUDA

This is CUDA implementation of the $NC^2$-algorithm for the word problem in hyperbolic groups. The algorithm was described in this paper.

  1. Hosting the model on a cloud server got too expensive, so the link is currently broken, but might become active in the future if I can find a spare computer to host it on.