Hello!

I’m Kyle, an undergraduate at Northeastern University. I like learning about the world, building things to improve it, and solving problems at the intersection of math and technology. Areas of interest include deep reinforcement learning, probability theory, and numerical optimization.

I enjoy all kinds of puzzles, rhythm games (primarily Arcaea; 13.13 ptt), riichi mahjong, and cooking.

Notable open-source work:

• Burn—deep learning framework (like PyTorch and TensorFlow) for Rust
• site—collection of programs for building this website
• haph—Rust library for flexible, high-performance compile-time hashmaps based on perfect hash functions
• AI2Fumo—browser extension replacing mentions of AI with Fumo
• fsbex—Rust library for parsing and extracting audio from the FMOD Sound Bank format commonly found in video games
• arkdata—asset fetcher for the video game Arknights; built on reverse-engineering private game API endpoints
• ArkDB—Arknights utility website for optimizing resource production and expenditure

I occasionally write about topics that interest me and projects I’ve worked on. Some highlights:

• Algorithms for calculating gacha probabilities (2025)