ML Accelerator Architect

Job Summary

The Waymo Compute Team is seeking a skilled engineer to join our multidisciplinary team. This role involves analyzing workloads, architecting efficient machine learning solutions, and collaborating with cross-functional teams to design and optimize high-performance computing platforms.

Responsibilities

• Analyze workloads and map them efficiently to hardware, proposing novel HW-friendly implementations and projecting performance
• Architect, simulate and design machine learning solutions for autonomous driving technology
• Work closely with compiler and model developers to influence engineering trade-offs and future model architectures
• Build scalable tools for simulator modeling and performance evaluation
• Interact with cross-functional engineering teams to identify opportunities and requirements

Requirements

• BS degree in Computer Science or Computer Engineering or similar relevant technical field, or equivalent practical experience
• 3+ years of experience designing/architecting complex, high performance architectures - CPUs, GPUs and/or ML accelerators - in the industry or through doctoral research
• 1+ years of experience with machine learning architectures, acceleration and model optimization
• Strong C++ programming and algorithmic problem solving skills

Preferred Qualifications

• 1+ years of experience modeling high performance architectures in cycle-aware simulators
• Track record of analyzing workloads and architecting, delivering novel HW+SW solutions to vastly improve performance, efficiency
• Familiarity with ML model architectures and their compute characteristics (bottlenecks, optimization opportunities)
• Experience with microarchitecture design (SystemVerilog or HLS)

Benefits

The expected base salary range for this full-time position across US locations is $170,000—$216,000 USD. Waymo employees are also eligible to participate in the Company’s discretionary annual bonus program, equity incentive plan, and generous benefits program, subject to eligibility requirements.