Projects

Useful and/or fun (hopefully) things I have been building on the weekends

Megaminx World Model Bench - symbolic puzzle-world RL environment

GT-Bench - verifiable game-theory reasoning benchmark

200loc: Interactive + complete step-by-step guide on how LLMs work

Fractal: The Infinite Curiosity Engine

Forecaster Arena - A new and uncontaminated LLM benchmark based on prediction markets

Towards Self-Improving Agents via the Traveling Salesman Problem - an experimental open environment

KineWatch - smart auto-speedup extension for YouTube

Hot or Slop - Turing test for generated images

Pose Compose - a next-gen image editor

Game of Life Survival Edition - single player take on Conway's mathematics

This Tourist Does Not Exist - an experiment with Generative Adversarial Networks