Developing vision capabilities for our onboard and offboard autonomous driving system.

Deep learning models fusing multiple sensors to predict 3D bounding box, vehicle trajectory, ground height and lanes.

GenAI synthetic data generation for domain randomization and expanding to ODDs such as snow tracks, dirt roads, forests etc.

Foundation models for autonomous driving built using multiple LiDAR, camera and RADAR sensors.

2D Segmentation for road, potholes, and obstacles.

Designed and deployed end-to-end ML system from active data collection to labeling to model training, evaluation and finally deployment.

Created a deep learning sensor fusion model to learn a unified semantic representation of the surroundings from which obstacles, vehicles, pedestrians, lanes etc could be identified.

Also developed a new inference engine to run 3D sparse convolutions and DL models with custom CUDA kernels on onboard systems under 40ms.

Developing AI infused features for Dynamics 365 Sales Insights.