All interviews
Google
Google
Google
Google
Mid
Optimize Google's GPU software stack from ML compiler cost models down to kernel-level performance
Join Google's Core ML organization to build optimizations for the latest generation of GPUs powering Google's ML infrastructure at massive scale. This role spans the full GPU software stack — from compiler cost model design to high-performance kernel tuning to cross-node model serving configuration.
Step into this interview
Free · a live voice mock calibrated to this exact role
What this interview tests
- Low-level GPU programming (CUDA/Triton/CUTLASS)
- GPU performance engineering and profiling
- GPU memory hierarchy and architecture
- ML compiler optimization (OpenXLA/MLIR)
- LLM deployment on accelerators
Common question themes
Walk through profiling and fixing a GPU kernel bottleneck
Memory-bound vs. compute-bound kernel diagnosis
How a compiler lowers/schedules ops for a GPU target
Tradeoffs in kernel tuning (occupancy, register pressure, tiling)
Cross-node serving considerations for large models
Related interviews
Senior
Software Engineer, Serverless Networking, Infrastructure
Google Clouddistributed systemsnetworking infrastructureserverless
See what it testsStaff
Staff Software Engineer, Mobile (Android), YouTube
AndroidMobileStaff EngineerTechnical Leadership
See what it testsAssociate
Software Engineer
Machine LearningGenAIDeepMindSoftware Engineer
See what it testsCohere
Mid
Site Reliability Engineer, Inference Infrastructure
SREKubernetesML infrastructureinference serving
See what it testsRoblox
Senior
Senior Software Engineer, App Performance
C++performance engineeringmobile/cross-platformgame engine
See what it testsDropbox
Staff
Staff Fullstack Software Engineer, Core Performance
Performance EngineeringReactStaff EngineerCross-Platform
See what it tests