Trainium (NKI) Kernel Expert
Mercor
$70 – $90/hour · as listed
This role involves evaluating and assessing Neuron Kernel Interface (NKI) development work for an advanced AI lab. You'll review code quality, verify correct implementation across hardware platforms, and ensure optimized performance on Trainium accelerators. Your main responsibility is providing detailed written feedback on kernel development tasks, checking for migration accuracy from CUDA, performance efficiency, and numerical correctness across different systems. This is AI infrastructure work requiring strong technical judgment and clear communication.
The position suits experienced developers with 2+ years of coding background who understand low-level optimization and hardware-specific constraints. You'll need advanced proficiency in kernel development, familiarity with accelerator architectures, and the ability to write precise technical assessments. This is a remote, hourly role in the US, available at 40 hours per week, paying $70–$90/hour based on experience.
From the listing: Mercor marketplace · hourly · 40h/week. Evaluate the quality, correctness, and hardware-appropriateness of Neuron Kernel Interface (NKI) development tasks used to train and evaluate a frontier AI lab's models. You'll assess CUDA→NKI migration fidelity, Trainium-specific performance-optimization quality, and cross-platform numerical-correctness standards — and provide clear, rubric-based written feedback. Basic Qualifications • 2+ years
Applications are completed on the listing’s own site. Pay is shown as posted by the source (“as listed”) or the aggregator’s estimate where marked — offers and availability change, and individual results vary.
Want help getting selected — and finding the best-paid work for your country? See how membership works →