GPU Kernel Expert

Put your GPU and accelerator kernel experience to work judging kernel development tasks. You check numerical correctness, fair benchmarking, task scope and whether code compiles and runs, and write clear feedback against a rubric.

What you will do

• Review GPU and accelerator kernel tasks for quality, correctness and completeness.
• Assess numerical correctness, performance-benchmarking fairness, task scoping and compilation and runtime validity.
• Work across many kernel task types.
• Write clear, rubric-based feedback on each task.

What you need

• 3+ years of hands-on experience developing, optimizing or verifying GPU or accelerator kernels in at least two of CUDA, Triton, NKI or Pallas (JAX).
• A strong grasp of numerical-correctness criteria for kernels, such as absolute, relative and ULP tolerances and choosing reference implementations.
• Experience with performance profiling and benchmarking (nsight, ncu, roofline analysis or framework-native profilers).
• Familiarity with common compilation and runtime failures: driver mismatches, OOM, launch-configuration errors, shape and stride mismatches, autotuning failures.
• Experience with at least three kernel task types: generation from a specification, translation or lowering across frameworks, migration between hardware targets, debugging, performance optimization or operator fusion.

Nice to have: experience with both NVIDIA GPU (CUDA, Triton) and custom accelerator (NKI, Pallas, TPU) ecosystems, a background in compiler engineering, MLIR or IR lowering, memory-hierarchy optimization skills, and contributions to kernel libraries such as cuBLAS, cuDNN, Triton community kernels or JAX and XLA custom calls.