top of page
INTERNSHIPS 1 - 2
Teaching sensors to think for themselves
— and understanding the business behind them
Two research internships with Prof. Kyusang Lee's Intelligence Sensing Group at UVA, plus a market-research internship at a mobility startup.
An artificial compound eye for 3D optical flow
NSF-supported · 40 hrs/week · Jul–Aug 2026
WHY
A flat camera sensor limits field of view and forces every pixel to travel to a separate processor. Insects solve this with a curved compound eye that senses and processes together — this project asks whether that can be built artificially.
HOW
I supported both sides: InGaAs photodiode film grown on graphene and transferred onto a domed array of 1P-1R smart pixels on hardware, and processing 16×16 optical-flow grids feeding into CSE-FSL, a split-computation framework, on software.
WHAT I DID
Split my time between the two sides of the build. On the hardware side, I supported testing of the photodiode and memory devices that make up each sensor pixel. On the software side, I processed the raw optical-flow output so the team could feed it into the split-learning framework that reads the sensor's data. Full methods and results are on the Research Projects page.
WHAT I TOOK FROM IT
I expected the hardware and software sides to line up neatly once each worked on its own. Instead, getting them to actually agree took longer than either side alone — a well-formed pixel array meant little until CSE-FSL could read its output correctly. Bridging that gap was less about one fix and more about going back and forth between materials and machine learning until they matched.
bottom of page