Built system fundamental code base and set up development environment. Conducted major integration efforts and set up vehicle environment to deploy software stacks. Re-wrote camera pipeline to improve performance, reaching stable 30FPS while reducing approximately 35% CPU consumption.

Designed software architecture and developed core infrastructure. Built debugging and integration tools to enhance collaboration and streamline development processes. Established standard practices within the repository, including coding convention, Docker, and initial CI/CD setup. Led main integration efforts and conducted major testing to ensure system reliability and performance. Led codebase profiling and optimization for all modules, proposed, and implemented GPU-based pipeline to improve inference performance of AI modules. Oversaw repository maintenance and merge request reviews to uphold performance and scalability standards.

Developed an SDK for integrating AI modules into Android applications for use in vehicles. Proposed and implemented a complete GPU pipeline with OpenGL and OpenCL backends, achieving approximately 2x faster inference while reducing CPU consumption by about 60%. Conducted profiling of the SDK to identify and optimize bottlenecks, ensuring efficient operation. Prepared in-vehicle hardware and managed the data collection process for deep learning modules.

Created the software foundation for the Motion Planning & Control module. Built a visualization application for algorithm debugging. Developed two autopilot features, including Blind Spot Warning and Forward Collision Warning. Conducted research on the Behavior Planning module for autonomous vehicles.