Technical Task Auditor — SWE-Bench Task Auditor

Meridial; United States; Remote; USD 83,000 - 165,000; Part time

Job summary

Meridial is seeking experienced technical experts for an independent contractor role focused on auditing tasks used to train and evaluate AI systems. You will be matched to a specialty such as GPU kernels, AWS Trainium, CVE/security, ML, AI-assisted workflows, AWS Serverless, Kubernetes, or SWE-Bench based on your experience.

Experts will assess task validity, realism, solvability, reproducibility, and support by tests and evaluation criteria, and provide actionable feedback.

Qualifications

• Experience auditing AI/ML systems or related software.
• Ability to assess accuracy, solvability, reproducibility, and test coverage.
• Willingness to complete targeted technical assessments during selection.

Responsibilities

• Assess technical tasks for accuracy, realism, solvability, and reproducibility.
• Provide clear, actionable feedback on issues identified with evaluation criteria and tests.

Skills

AI auditing

SWE-Bench

Machine learning

GPU kernels

Kubernetes

AWS Trainium

Security testing

AI-assisted developer workflows

Tools

Kubernetes

AWS

Job description

We are seeking experienced technical experts to contribute to an independent contractor opportunity focused on auditing tasks used to train and evaluate AI systems.
Experts will be matched to one specialty based on their experience:

• GPU kernels
• AWS Trainium and NKI
• CVE and application security
• Machine learning
• AI-assisted developer workflows
• AWS Serverless and Infrastructure as Code
• Kubernetes
• SWE-Bench

Experts will assess whether technical tasks are accurate, realistic, solvable, reproducible, and supported by reliable tests and evaluation criteria. They will also provide clear, actionable feedback on any issues identified.
Expertise across all eight specialties is not required. Interested experts will complete a technical assessment related to their area of expertise as part of the selection process.

Required assessments

• SWE-Bench Task Auditor Assessment (50 min)