Training machine learning models for self driving vehicles.

Developed a convolutional neural network for lane detection using camera & lidar. Created training dataset by cleaning and auto-labeling data logged from Kodiak’s truck fleet. Trained the detector in Pytorch and used TensorRT, OpenCV, and C++ to run it in real time. Integrated the detector with internal visualization and simulation tools. Deployed the detector to the entire fleet.

Modified the Skydio drone's object tracking abilities to use a dynamically changing camera model, allowing users to adjust the camera's zoom level in real time. Shipped feature to customer drones.

Developed an autonomous drone capable of automatically finding and disinfecting door handles. Physically designed, assembled, and calibrated the drone. Built the autonomy stack using C++, ROS, SLAM, a YOLO detector, RTT* path planning, trajectory generation, and a state machine.

Selected to be a teaching assistant for introductory CS classes. Taught a weekly section and helped students with their Java & C++ assignments.