- Designed and built the C++ flight software stack for a fleet of 50+ fully autonomous drones, enabling 40,000+ inspections so far with industry-leading 15 min/turbine efficiency. - Engineered efficient viewpoint-planning and camera perception algorithms, achieving >99% accuracy in detecting 1 cm-level damages under dynamic conditions. - Developed a low-latency, graph-based time synchronization library, enabling microsecond-accurate data association across 7+ high frequency sensor streams (>200 Hz). - Led software innovations to expand inspections into GPS-degraded floating offshore wind turbines, overcoming localization and mapping challenges in dynamic maritime environments. - Built LiDAR perception systems to unlock real-time semantic segmentation and safe navigation at >4 m/s. - Developed a decentralized multi-agent mission planner, resolving 10+ competing objectives at 100 Hz.

Implemented evolutionary algorithms for combined task and motion planning in automated construction.

Developed numerical models for design optimization and verification of high-tension lines with dampers.

Led a subsystem of 6 people responsible for the underbody suspension and PID-controlled eddy-braking for the levitating 'OrcaPod'. One of 24 global finalists in the SpaceX Hyperloop competition held at SpaceX HQ, CA.