GPU Kernel Expert - AI Specialist

Obsidian; San Francisco (CA); On-site; USD 120,000 - 180,000; Full time

Job summary

Obsidian is seeking a skilled reviewer to evaluate GPU/accelerator kernel development tasks used to train frontier AI models. You will assess numerical correctness, performance benchmarks, and task scoping, ensuring compilation and runtime validity across diverse kernels.

The role requires 3+ years in CUDA, Triton, NKI, or Pallas, plus experience with profiling tools and handling common failure modes. This is a research-oriented, full-time position in San Francisco.

Qualifications

• 3+ years of hands-on experience developing, optimizing, or verifying GPU/accelerator kernels in two or more of CUDA, Triton, NKI or Pallas (JAX).
• Strong understanding of numerical-correctness criteria for kernels (absolute/relative/ULP tolerances, reference-improvement selection).
• Experience with performance profiling and benchmarking (Nsight, ncu, roofline analysis or framework profilers).
• Familiarity with common compilation and runtime failure modes (driver mismatches, OOM, launch-configuration errors, shape/stride mismatches).
• Experience with at least three kernel task types: generation from specification, translation/lowering, migration across hardware targets, debugging, optimization, or operator fusion

Responsibilities

• Evaluate the quality, correctness, and completeness of GPU/accelerator kernel development tasks.
• Provide rubric-based written feedback to train and assess a frontier AI lab's models.
• Assess compilation/runtime validity across diverse kernel task types.
• Produce actionable recommendations to improve task scoping and benchmarking fairness

Skills

CUDA

Triton

NKI

Pallas (JAX)

Tools

Nsight

ncu

Roofline

Job description

Evaluate the quality, correctness, and completeness of GPU/accelerator kernel development tasks used to train and evaluate a frontier AI lab's models. You'll assess numerical correctness, performance-benchmarking fairness, task scoping, and compilation/runtime validity across diverse kernel task types — and provide clear, rubric-based written feedback.

Basic Qualifications

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

Preferred Qualifications

• Experience across both NVIDIA GPU (CUDA/Triton) and custom-accelerator (NKI/Pallas/TPU) ecosystems
• Background in compiler engineering, MLIR, or intermediate-representation lowering
• Understanding of memory-hierarchy optimization (shared-memory tiling, register pressure, bank conflicts, coalescing patterns)
• Contributions to kernel libraries (cuBLAS, cuDNN, Triton community kernels, JAX/XLA custom calls)