• Leveraged PyTorch and YOLOv8 classification model to extract landing pad locations from real-time video• Engineered the decision-making algorithm by integrating telemetry data, predicted landing pad coordinates, and ground-station commands to ensure precise navigation and successful landings• Interfaced with the ArduPilot flight controller using MAVLink to monitor the drone’s status and transmit commands

• Debugged, reviewed, and submitted as a member of the MLCommons MLPerf Inference Working Group, focused on creating a set of fair AI benchmarks for edge computing devices• Developed board support packages (BSPs) for Nvidia Jetson carrier boards through kernel driver development, device tree creation, and shell scripting automation, adding robust support for GMSL/MIPI cameras and I2C/SPI devices• Implemented disk encryption and OTA updates, enhancing security and serviceability of devices deployed in the field

• Co-created DATASCI 3VG4, an upper-year university course on physics simulations, with Dr. James Wadsley• Developed force and impulse-based, real time, rigid body physics simulations using Panda3D and Python• Designed and implemented algorithms to triangulate, render, and detect collisions of prisms and ellipsoids