Staff Software Engineer, Compute Reliability

The Planner/Perception Reliability team's goal is to build out architectures, tools, and workflows to prevent, identify, and guide fixes of reliability and software integrity issues.

We focus the organization on reliability and are accountable for the onboard software health while enabling high development velocity.

- Drive the design and development of instrumentation and onboard architecture to proactively prevent, detect, and debug reliability issues across current and future Waymo compute platforms.

- Design and implement long-term strategies to systematically mitigate compute hardware reliability and numerical stability issues across heterogeneous platforms (CPU, TPU, GPU), creating scalable tools to automate triage, debugging, and resolution.

- Lead cross-functional initiatives across Waymo, Alphabet, and external partners to architect proactive solutions that prevent issues originating from compilers (LLVM, XLA, JAX), optimization (AutoFDO), or model changes.

- Architect frameworks and diagnostics to proactively identify and eliminate complex software and firmware faults, including deadlocks, memory corruption, race conditions, memory leaks, or tail latency spikes.

- Influence the design and tooling of next-generation hardware platforms, serving as a technical authority to ensure a coherent, forward-looking reliability architecture.

- BS/MS in Comp Sci, EE, Robotics, Physics, Math, or related field (or equivalent experience).

- Experience with CUDA/GPU/TPU acceleration and model reliability or optimization.

- Experience with writing GPU kernels and evaluation/debugging of GPU workloads.

- MS or PhD in Computer Science, Robotics, similar technical field of study, or equivalent practical experience.

- Experience with autonomous vehicles (L4) or ADAS systems (L2/L3).