Curriculum Vitae
Computer vision research scientist developing novel solutions for human sensing
Lead scientist for eye tracking at Meta Reality Labs’ AR/VR research, with experience spanning the entire pipeline from early concept to shipped models in production
Deep coder solving technical problems that unblock entire teams
Work Experience
Technical lead scientist for research team of 20 scientists and engineers, overseeing all eye tracking research for Meta
Built synthetic data infrastructure leveraging 3D reconstructed scans of human eyes, reducing data required for training ML models by 80%; enabled zero-real-data models for rapid iteration of prototype eye tracking hardware
Drove AI adoption effort for org of 200 engineers, resulting in increased velocity of code through testing and rigorous specification of AI-generated code
Extensive hiring expertise: led committee to standardize computer vision design interview questions for entire Meta AR/VR org; conducted over 550 interviews for scientists and engineers
Designed and built multiple 3D reconstruction algorithms based on Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS) for human eye capture to handle dynamic eye motion, relightability, and optical effect simulation
Built multi-camera capture rig for 3D reconstruction of human eyes; led team to capture 500+ scans of real human eyes
Developed novel sensing technology using combination of applied physics and computer vision to prototype eye tracking hardware using polarization, event cameras, waveguide imaging, self-mixing interferometry, scanning-laser MEMS, and laser speckle
Designed and built novel display calibration robot implementing a virtual Shack-Hartmann wavefront sensor for characterization of distortion and focus of VR displays
Implemented real-time rendering and control software for mechanical varifocal headset prototype (Half Dome) resulting in a demo at Oculus Connect; published study evaluating improved user comfort from varifocal display
Mentored 13 PhD research interns
14 publications in refereed conferences and journals, 39+ patents
Designed and implemented optimization algorithms for computer vision inference, using convex programming and duality
Developed theoretical framework for continuous Markov Random Field inference
Education
Technologies
Python (11 years), C++ (18 years), Claude, pytorch, OpenCV, MATLAB, ceres-solver, matplotlib, jupyter, mercurial, git