Remote Trainium NKI Kernel Expert: CUDA Migrations

Appsierra Group; United States; On-site; USD 146,000 - 187,000; Full time

Job summary

Trainium is seeking a Kernel Expert to evaluate NKI kernel tasks and ensure hardware-appropriate, optimized implementations. This fully remote independent contractor role focuses on high‑fidelity CUDA-to-NKI migrations and Trainium-specific performance tuning.

You will review tile-based computation, memory hierarchies (SBUF/PSUM/HBM), and DMA orchestration, delivering rubric-based feedback on kernel solutions.

Qualifications

• 2+ years hands-on kernel development or optimization with NKI.

Responsibilities

• Evaluate NKI kernel tasks for quality, correctness, completeness, and hardware appropriateness.

Skills

NKI kernel experience

CUDA-to-NKI migrations

Trainium hardware knowledge

Performance profiling

DMA orchestration

Tile-based computation

SBUF/PSUM/HBM memory hierarchy

Cross-platform numerical correctness

Tools

AWS Neuron SDK

Neuron Compiler internals

CUDA kernel development

Triton kernel development

Job description

Trainium is seeking a Kernel Expert to evaluate NKI kernel tasks and ensure hardware-appropriate, optimized implementations. This fully remote independent contractor role focuses on high‑fidelity CUDA-to-NKI migrations and Trainium-specific performance tuning.

You will review tile-based computation, memory hierarchies (SBUF/PSUM/HBM), and DMA orchestration, delivering rubric-based feedback on kernel solutions.