Teams at Zoox

Optimize GPU performance for Zoox’s motion planner, including profiling GPU hotspots, adapting software to multiple GPU architectures, reducing ML model latency and memory usage, and developing or optimizing CUDA kernels. Serve as a CUDA and GPU performance subject matter expert while working with complex multiprocess robotic systems in a Linux and C++ environment.

Optimize GPU-based motion-planning algorithms for latency, memory usage, and resource utilization. Analyze performance metrics, identify GPU hotspots, adapt the planner across GPU architectures, optimize machine-learning models, and support engineers as a CUDA and GPU-performance subject matter expert. The role involves C++ development in large codebases, Linux environments, and debugging and profiling complex multiprocess robotic systems.

Develop distributed computing algorithms, performance metrics, and analysis tools for Zoox’s Collision Avoidance System. Analyze petabytes of urban driving data, evaluate system performance, and collaborate with CAS Perception, Planning, Systems Design, Mission Assurance, and QA engineers to define autonomous-driving validation metrics and plans.