GPU Kernel Quality Auditor for AI Task Validation

Remote DXB; United States; Remote; USD 165,000 - 358,000; Full time

Job summary

Remote DXB is seeking a specialist to audit GPU kernel tasks used to train and evaluate AI systems. You will assess technical accuracy, realism, solvability, reproducibility, and the strength of tests and criteria for each task.

The role requires deep CUDA/Triton programming knowledge, parallel computing expertise, and a strong ability to identify memory issues and optimization opportunities through rigorous analysis and feedback.

Qualifications

• Demonstrable professional experience and deep technical knowledge in GPU Kernel development (including CUDA/Triton programming, parallel computing, and hardware-level performance optimization).
• Strong analytical and problem-solving skills to rigorously test, troubleshoot, and evaluate complex technical scenarios.

Responsibilities

• Assess whether GPU kernel tasks are technically accurate, realistic, solvable, reproducible, and supported by reliable tests and evaluation criteria.
• Provide clear, actionable feedback on memory allocation bugs, execution bottlenecks, or parallel computing logic errors within the tasks.

Skills

GPU Kernel development

CUDA/Triton programming

Parallel computing

Hardware-level performance

Job description

Remote DXB is seeking a specialist to audit GPU kernel tasks used to train and evaluate AI systems. You will assess technical accuracy, realism, solvability, reproducibility, and the strength of tests and criteria for each task.

The role requires deep CUDA/Triton programming knowledge, parallel computing expertise, and a strong ability to identify memory issues and optimization opportunities through rigorous analysis and feedback.