5 months ago
Boston, MA, USA +3 moreMid Level
H1B Sponsor
Base Salary
$168k - $239k/yr
Responsibilities
- Build real-time instrumentation for CPU, GPU, latency, and memory performance monitoring.
- Develop offline benchmarking frameworks, tools, and scripts to evaluate performance at scale in CI and vehicle environments.
- Establish performance budgets for next-generation compute architectures.
- Analyze performance metrics to identify GPU hotspots and root causes.
- Propose and co-implement performance solutions with component teams.
- Help bring serial algorithms to the GPU to maximize compute utilization and improve latency.
- Contribute to a middleware framework that encourages efficient CPU and GPU code development.
Requirements
- BS in computer science or a related field.
- At least 3 years of experience.
- Strong knowledge of CUDA and recent GPU microarchitectures such as Ampere and Blackwell.
- Experience debugging and optimizing GPU kernels with tools such as Nsight.
- Strong C++ knowledge and experience working in large codebases.
- Comfort working in Linux development environments.
- Experience developing, debugging, and profiling complex multiprocess systems such as robotic systems or game engines.
- Preferred experience with real-time GPU kernel development, PTX-level programming, CPU SIMD instructions such as AVX intrinsics, and custom CUDA layers using TensorRT and XLA.
- Preferred hands-on experience with machine learning model optimization, including post-training quantization and layer pruning, or GPU kernel tuning with OpenGL, CUDA, ROCm, or similar technologies.
- Proficiency with SQL, Databricks, Looker, or other business intelligence tools is preferred.
About Zoox
Zoox is transforming mobility-as-a-service by developing a fully autonomous, purpose-built fleet designed for AI to drive and humans to enjoy.