Adversarial Risk Evaluator AI Trainer, [$20–$30]/hour

Role Overview

Project consultant (AI Trainer) to evaluate AI-generated prompt–response pairs and improve LLM performance for trust & safety and content quality. This role is part-time, project-based, and U.S.-based experts only.

Responsibilities

• Review prompt–response pairs produced by AI systems
• Assign relevance scores using defined rating scales
• Evaluate and classify safety risks across categories such as:
  • hate
  • sexual content
  • violence
  • self-harm
• Apply severity ratings according to provided policy guidelines
• Provide brief, clear written justifications for each evaluation
• Maintain consistency across large volumes of tasks
• Meet quality standards and incorporate feedback from QA/calibration

Qualifications

• Experience in trust & safety, content moderation, QA, or AI evaluation (preferred)
• Strong written communication and judgment
• Ability to follow detailed guidelines and apply structured rubrics consistently
• Comfortable reviewing sensitive/explicit content
• High attention to detail and ability to work independently in a remote environment