Stealth AI startup

Led a cross-functional team that built a voxel-based occupancy network for short-range 3D perception in autonomous vehicles, fusing Lidar, Radar, and Camera data into a unified world model. Nearly eliminated false-positive detections in adverse weather. Managed the full development lifecycle across a team of 8+ engineers, from autolabeling pipelines to large-scale model training and production deployment.

Built state estimation and perception systems for agricultural robots. Acquired by Zoox.

Developed robust place recognition and localization algorithms for large-scale augmented reality on mobile devices.

Developed real-time, dense 3D simultaneous localization and mapping (SLAM) using passive monocular cameras for use on micro-aerial vehicles (MAVs).