I build ML systems, compilers, and autonomous systems.
Currently studying Applied Mathematics/ML at the University of Waterloo, leading prediction work at WATonomous, and doing research at Waterloo's Vision and Image Processing Lab.
-
comma Controls Built GPU-accelerated optimization and MPC tooling for comma's Controls Challenge, then derived the benchmark's closed-form optimum and reached a 6.880 leaderboard score. The project documents the optimization pipeline, parity testing, ablations, negative results, and benchmark analysis.
-
WATonomous Prediction Working on multi-modal trajectory prediction for autonomous vehicles using C++, Python, ROS2, HD-map lanelets, ONNX, and TensorRT, including model integration and production prediction infrastructure.
-
3DGS → Physics Simulation · VIP Lab Working on pipelines that convert 3D Gaussian Splatting scenes into anisotropic particles for physics simulation, including C++ geometry tooling and large point-cloud processing.
C++ · Python · PyTorch · CUDA · ROS2 · TensorRT · ONNX · Triton · CMake · Linux · Docker · TypeScript · React


