Projects
- Computer Vision
At-Home VAR
Recreating the World Cup's offside VAR with a single broadcast camera instead of a whole stadium of them - it runs the full computer-vision pipeline in your browser to call onside or offside.
- Computer Vision
CourtVision
A coach uploads a tennis match video and my computer-vision pipeline breaks down every shot - winners, errors, serve placement, rally length, and court-zone heatmaps - turning raw footage into real analytics.
- Computer Vision
Athletic Training Portal
I'm the CTO of ATP, a sports biomechanics platform that measures an athlete's range of motion and symmetry from nothing but phone video - we're working with real teams already and it's been incredible to see it give coaches data they could never get before.
- Predictive
CHALK
CHALK predicts winners and scores across six leagues without ever looking at the betting line, which turns the model into a clean control - the gap between it and the market becomes a map of everything the crowd knows that I don't.
- Other Software
Munchr
Munchr helps Duke students find snacks across campus vending machines, but the cooler half is the flip side - every search becomes demand data showing operators what people wanted but couldn't find.
- Tools
Media Converter
Drop in a whole folder of images, audio, video, or docs, pick a target format for each type, and it converts them all right in your browser and hands them back mirroring your exact folder structure.
- Tools
Audio Stitcher
Upload a few audio clips and it stitches them into one seamless mix, with the beat-matching, crossfades, and loudness mastering all running right in your browser through WebAssembly.
- Computational Biology
DynaLab
DynaLab simulates how proteins fold, flex, and unfold under mechanical force, reaching the microsecond-to-millisecond timescales where the biologically interesting stuff actually happens - all from a guided web app instead of a cluster and a pile of command-line tools.
- Computational Biology
Data Augmenter
A data-augmentation tool that uses a ControlNet diffusion model to turn simulated biological patterns into realistic synthetic training data, basically closing the gap between what we can simulate and what we actually see in experiments.
No projects in this category yet.