Software Engineer, GPU Performance

Google Google · Big Tech · Sunnyvale, CA +1

Software Engineer focused on optimizing GPU performance for Google's AI/ML infrastructure, impacting billions of users. Responsibilities include low-level GPU programming, performance engineering, and influencing the GPU software ecosystem and fleet deployment.

What you'd actually do

  1. Build optimizations for the latest generation of GPUs that power Google’s most critical products and services, impacting billions of users worldwide.
  2. Address the most challenging performance bottlenecks through Google’s unparalleled access to the latest generation of GPUs, tooling, and a decade of experience building AI accelerators.
  3. Drive optimizations across Google’s GPU software stack from ML compiler cost model design to optimizing high performance GPU kernels to cross node model serving configurations.
  4. Influence the technical direction of the GPU software ecosystem at Google by collaborating with ML, compiler design, and systems architecture.
  5. Influence the deployment of Google’s GPU fleet by working with various product teams across Google.

Skills

Required

  • software development
  • low-level GPU programming (CUDA, Triton, CUTLASS, etc.)
  • performance engineering techniques
  • modern GPU architectures (NVIDIA, AMD, or other AI accelerators)
  • memory hierarchies
  • performance bottlenecks

Nice to have

  • data structures
  • algorithms
  • compiler optimization
  • code generation
  • runtime systems for GPU architectures (OpenXLA, MLIR, Triton, etc.)
  • Large Language Models (LLMs)
  • deployment on AI accelerators

What the JD emphasized

  • low-level GPU programming
  • performance engineering
  • modern GPU architectures
  • GPU software stack
  • ML compiler cost model design
  • optimizing high performance GPU kernels
  • cross node model serving configurations

Other signals

  • GPU performance optimization
  • ML infrastructure
  • AI accelerators