Fast, accurate optical flows on mobile GPUs.
A design and implement an optical flow algorithm targeted at a 7.5W TDP mobile system, capable of processing 4K resolution frames in realtime, 24 fps.
We achieve a 1,250x speedup over OpenCV by using optimized GPU kernels, multithreading, and fine-grained memory management.
Award winner at the 2017 Carnegie Mellon Parallelism competition.
Vector-based handwriting recognition engine.
Machine learning driven touchpad calculator, using a novel, vector-based feature detection algorithm.
Awarded 4th place at the Fundamentals of Programming and Computer Science showcase.
Beautiful resumes. No headaches.
Write a yaml file. Outputs a pdf. No Adobe products needed. Looks better too.