Teams at Zoox

Design and optimize on-vehicle perception inference: allocate CPU/GPU resources, compress and quantize large multimodal models, build TensorRT model conversion pipelines, and deliver production-quality low-latency C++/CUDA inference for power- and thermal-constrained vehicle SoCs.

Design, build, and operate Zoox's HPC platform supporting thousands of concurrent jobs. Modernize infrastructure (Ray, SLURM, Kubernetes, AWS), improve scheduling, autoscaling, multi-region orchestration, capacity planning, profiling, and developer tooling. Lead cross-team initiatives and mentor junior engineers.

Own and enhance Zoox's Bazel-based build ecosystem for a large polyglot monorepo: optimize build pipelines, develop Starlark rules and tooling, integrate quality tools, and build dashboards to improve developer velocity and reliability.