Sarah Pan, a research fellow at Answer.AI and MIT first-year student, shares her journey from learning AI in middle school via fast.ai to publishing at NeurIPS and building WebGPU Puzzles — a browser-based, beginner-friendly GPU programming education tool. The project is modeled after Sasha Rush's CUDA Puzzles and uses WebGPU and gpu.cpp to teach parallel computing fundamentals without requiring C++ expertise. Sarah describes her motivation to make GPU programming accessible, drawing on her own experience learning it from scratch in a matter of weeks.
29 Impressions