Unknown Company

Trainium NKI Kernel Developer for AI Model Evaluation

West Menlo Park, California • Posted Today
Remote Full Time technology

Help raise the quality of frontier AI training and evaluation by reviewing Neuron Kernel Interface, NKI, development tasks for correctness, hardware fit, and performance on AWS Trainium and Inferentia2.

Role Overview

You will evaluate CUDA-to-NKI migration fidelity, Trainium-specific optimization quality, and cross-platform numerical correctness. Your assessments will provide clear, rubric-based written feedback on kernel development tasks.

Key Responsibilities
  • Review NKI kernel tasks for quality, correctness, and appropriate use of Trainium hardware.
  • Assess the fidelity of CUDA-to-NKI migrations.
  • Evaluate Trainium-specific performance optimizations, including NeuronCore pipeline utilization, tensor-engine throughput, and memory-bandwidth bottlenecks.
  • Assess cross-platform numerical-correctness standards, including GPU and Trainium accumulation order, rounding behavior, and mixed-precision semantics.
  • Deliver clear written evaluations using established rubrics.
Qualifications
  • At least 2 years of hands-on experience developing or optimizing NKI kernels for AWS Trainium or Inferentia2 hardware.
  • Strong knowledge of tile-based computation, SBUF, PSUM, and HBM memory-hierarchy management, partition-dimension constraints, and DMA orchestration.
  • Demonstrated experience evaluating CUDA-to-NKI migration quality.
  • Experience defining or assessing cross-platform numerical-correctness standards.
Preferred Qualifications
  • Experience with the AWS Neuron SDK, Neuron Compiler internals, or contributions to NKI kernel libraries.
  • Prior CUDA or Triton kernel-development experience.
  • Familiarity with NeuronCore-v2 architecture, on-chip SRAM topology, and supported FP32, BF16, FP8, and INT8 data types.
  • Experience benchmarking machine-learning training workloads on Trn1 or Trn2 instances.
Work Terms
  • Remote role, available to candidates based in the United States.
  • Hourly engagement.
Compensation

$70 to $90 per hour.

Trainium NKI Kernel Developer for AI Model Evaluation in West Menlo Park at Unknown Company

This position is listed as full time and able to be worked remotely.

Back to Job Search