CS @ University of Waterloo
Website · LinkedIn · Résumé · Email
I work on high-performance computing and efficient AI, from accelerating 3D reconstruction to building concurrent data structures.
- Research Intern @ Huawei Noah's Ark Lab — efficient spatial foundation models.
- Incoming SWE Intern @ DeepInfra — LLM inference acceleration, Jan-Apr 2027.
- Previously at Waterloo's Multicore Lab and Wolfram Research, working on concurrent data structures and parallel cellular automata.
HonkPack · HackMIT 2026
Phone video → per-object 3D reconstruction → custom 3D bin packing. Built with SAM3, Depth Anything 3, and ShapeR. Project page ↗
Rewrite · TreeHacks 2026
Interactive, AI-generated worlds in VR. Scaled Hunyuan WorldPlay inference from 1 to 12 FPS with tensor parallelism across 4× H200 GPUs. Demo & write-up ↗
ADaptiv · CMU NexHacks 2026
End-to-end video object replacement: Gemini for video understanding, SAM3 for pixel-level masking, and Wan2.1 for generation in ComfyUI. Optimized for a 3× speedup on an RTX 4090. Demo & write-up ↗
2× CCO silver medalist · USACO Platinum
I now help write the contests, too: CCO 2026 · Day 1, Problem 3.


