Brain-Computer Interface Wheelchair
published · CBS
A close friend's grandfather was losing the ability to walk, so a few of us built a BCI
wheelchair prototype in his honor. A real-time EEG pipeline in C++ and Python on a
Raspberry Pi, turning noisy signals into movement, with ML classification tuned for
real-world noise. Presented at the Advancing Healthcare Innovation Summit, featured on
CBS, and published in the Journal of Innovation in Digital Health, Diagnostics, and
Biomarkers.
It is the project I am proudest of, because it started with somebody specific rather
than with a spec.
- C++
- Python
- Raspberry Pi
- NumPy
- SciPy
- OpenBCI
Thermal-Imaging Friendly-Fire Avoidance System
US20240085134A1
A patented real-time system that detects human silhouettes in live infrared camera feeds.
The hard part was never detection; it was detection fast enough to matter, since in a
safety system a correct answer that arrives late is worse than useless. I profiled the
pipeline end to end and optimized the hot path.
- Python
- OpenCV
- Embedded Linux
Podman
open-source contributor
container ancestry
Fixed a bug in Podman's container ancestry logic, where a container reported the wrong
ancestor image under certain conditions. It is also where my love of nvim started:
jumping to the definition of a type and landing exactly where you meant to is a small
thing that never stopped being satisfying.
Preference-Based News App
team lead
Texas Convergent
Led a student engineering team building a short-form news app with a personalized feed
driven by the topics each reader picks.