2 hours ago
Base Salary
$251k - $310k/yr
Responsibilities
- Analyze model architectures and identify training and inference bottlenecks involving memory bandwidth, compute, and communication.
- Develop and apply quantization, pruning, knowledge distillation, and efficient attention techniques.
- Optimize model code for TPU and GPU accelerators using compiler features and low-level libraries.
- Evaluate data, tensor, pipeline, and expert parallelism strategies to improve scalability and efficiency.
- Design and implement low-latency, high-throughput serving solutions for generative models.
- Optimize training pipelines and build tools for model performance analysis, profiling, and debugging.
Requirements
- MS or PhD in Computer Science, Machine Learning, Robotics, or a related field.
- 5+ years of experience with deep learning architectures, algorithms, and optimization techniques, especially Transformers, Diffusion Models, and MoEs.
- Proficiency in JAX and Flax, with potential experience in TensorFlow or PyTorch.
- Expertise with ML workload profiling tools such as XProf, Perfetto, and NVIDIA Nsight.
- Hands-on experience with quantization, pruning, distillation, and other model-compression methods.
- Strong Python programming skills and potentially C++, with software development best-practice experience.
- Preferred: knowledge of TPU and GPU architectures, XLA and ML compiler behavior, multi-device and multi-machine training and serving, and contributions to JAX, Gemax, or XManager.
Benefits
- Eligibility for Waymo’s discretionary annual bonus program, equity incentive plan, and generous Company benefits program, subject to eligibility requirements.
- Full-time position with salary ranges varying by US location; remote-location salary details may be provided during the hiring process.
Categories
About Waymo
Waymo develops the Waymo Driver, a full-stack autonomous driving system, and operates the Waymo One robotaxi service that offers paid, driverless rides in select U.S. cities. The company monetizes through ride-hailing fares and partnerships integrating its technology on multiple vehicle platforms. Founded in 2009 as Google’s self-driving project, Waymo is headquartered in Mountain View and is an Alphabet subsidiary, with over 10 million rider-only trips and 100+ million autonomous miles on public roads.