Trainium (NKI) Kernel Expert
MercorLegit · not rated yetPosted on 25 Aug 2026
Pay: $70–90/hr
- Who can apply
- Open only to people in United States
- Hours
- 40 hrs/week · Contract
- Where
- Remote, work from home
- Job type
- Expert (SME) review
- Openings
- 3 openings
At a glance: A contract, remote expert (SME) review job with Mercor. It pays $70–90/hr for 40 hours a week. Only people in United States can apply.
Ongoing · listed since 25 Aug 2026
Mercor keeps this listing open to recruit all year. It is still open, but it isn’t a new opening.
What happens after you click Apply
- You leave AI Trainer HQ and go to Mercor’s own site.
- Upload your résumé on work.mercor.com
- Confirm your work authorization
- Sit the AI interview, which takes about 20 minutes
- Complete any written assessment attached to the listing
- On receiving an offer, sign it, configure payouts and verify your identity
How this pay compares
This job pays more than 73% of the 204 other science (PhD) jobs on the board that state pay; 26% pay more.
- This job
- $80/hrMiddle of $70–90/hr
- Median on Mercor
- $80/hrFrom 295 other paid jobs
- Median in science (PhD)
- $65/hrFrom 204 other paid jobs
Requirements
- Qualifications
- Not specified in the listing
- Languages
- English
- Equipment
- Computer with a reliable internet connection
About this job
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 of hands-on experience developing or optimizing kernels using the Neuron Kernel Interface (NKI) targeting AWS Trainium/Inferentia2 hardware
• Strong understanding of NKI-specific development patterns: tile-based computation, SBUF/PSUM/HBM memory-hierarchy management, partition-dimension constraints, and DMA orchestration
• Demonstrated experience assessing CUDA→NKI migration quality
• Familiarity with Trainium-specific performance profiling (NeuronCore pipeline utilization, tensor-engine throughput, memory-bandwidth bottlenecks)
• Experience defining or evaluating cross-platform numerical-correctness standards (GPU vs Trainium accumulation order, rounding behavior, mixed-precision semantics)
Preferred Qualifications
• Direct experience with AWS Neuron SDK, Neuron Compiler internals, or contributions to NKI kernel libraries
• Prior CUDA or Triton kernel development
• Familiarity with Trainium hardware specifications (NeuronCore-v2 architecture, on-chip SRAM topology, supported data types: FP32/BF16/FP8/INT8)
• Experience benchmarking ML training workloads on Trn1/Trn2 instances
Listing history
- First seen on our board
- (9 days ago)
- Posted by Mercor
- Last checked
- Source
- Mercor
We check every live job against its source at least every 24 hours. How we check jobs
