Ph.D. candidate, Electrical & Computer Engineering · Purdue University
I work on computational imaging, co-designing optics and algorithms. My research combines metasurface and refractive optics with diffusion and flow-matching models to build compact cameras that see more with less. I am advised by Prof. Stanley Chan and Prof. Qi Guo.
I completed my B.Sc. in Computer Engineering at the University of Peradeniya in 2020. Outside research I enjoy competitive programming and music.
A compact refractive-metasurface telephoto camera paired with a one-step diffusion model that fuses a structure image and a color cue, reaching a telephoto ratio of 0.44 with continuous 20–50 mm zoom.
A diffusion-based algorithm that corrects the optical aberrations of metalens cameras.
The first generative multi-frame RAW-to-RGB super-resolution pipeline, using foundation-model image priors to recover sub-pixel detail from bursts.
A CNN for intrinsic image decomposition trained without supervision, using physics-based parameters derived from the image.
A forecasting framework for COVID-19 case trends at sub-national resolution.
A study of the practical limits of learning models in dynamic wireless environments (Nokia Bell Labs).
Low-light enhancement using paired and unpaired data with cycle consistency.
Foreground estimation through background cancellation using classical spatio-temporal signal processing.
Sleep apnea identification in infants from a video feed.
Extreme low-light image enhancement using flow matching models.
Efficient burst RAW super-resolution with foundation image models (GenMFSR, CVPRW 2026).
Computational imaging with metasurface optics (MetaTele, ICASSP 2026). Advised by Prof. Stanley Chan and Prof. Qi Guo.
AI framework for threat assessment and containment of COVID-19 and future epidemics.
Computer vision for low-light images, including intrinsic image decomposition.
Deep learning for the physical layer of mobile networks under latency constraints.
