At Driveblocks, I develop Python and C++ components for camera- and LiDAR-centric perception and multimodal fusion, working with camera, radar, ego-motion, and vehicle-state data. I also evaluate sensor-fusion models and temporal consistency for safety-critical deployment.
At Plus AI, I built real-time uncertainty-aware out-of-distribution detection for multimodal deep learning systems and developed tools for robustness evaluation and failure analysis. Earlier, I developed transformer-based forecasting models at General Motors Cruise and worked on trajectory prediction, multi-object tracking, and sensor fusion at Brightskies Technologies.
