Remote AI Safety Evaluator for RLHF & Policy

mercor; Portugal; Remote; EUR 74.000 - 86.000; Vollzeit

Zusammenfassung

Mercor is seeking an AI Safety Practitioner to evaluate AI-generated responses for safety, factual accuracy, policy compliance, and overall quality in a fully remote role. The position offers contract terms and the ability to contribute to cutting-edge AI safety benchmarking.

You will review content involving misinformation, political persuasion, self-harm, violence, cyber, biosecurity, and other sensitive domains; apply evaluation rubrics for RLHF and SFT; identify unsafe outputs; and provide

Qualifikationen

• Bachelor’s degree or higher in a related field (e.g., Journalism, Psychology, CS, Public Policy).
• 5+ years of professional experience in AI Safety, Trust & Safety, journalism, policy, or related field.
• Excellent written English, critical thinking and analytical reasoning.
• Ability to evaluate nuanced, policy-sensitive scenarios.

Aufgaben

• Evaluate AI-generated responses for safety, factual accuracy, and policy compliance.
• Review content in misinformation, political persuasion, self-harm, violence, cyber, and biosecurity.
• Apply and refine rubrics for RLHF, SFT, and AI safety benchmarking.
• Identify unsafe outputs, hallucinations, reasoning failures, and policy violations.
• Provide structured feedback to improve model alignment and safety.
• Collaborate with AI researchers and safety teams on ongoing evaluation initiatives.

Kenntnisse

Excellent written English

Critical thinking

Analytical reasoning

Ausbildung

Bachelor's degree or higher in related discipline

Jobbeschreibung

Mercor is seeking an AI Safety Practitioner to evaluate AI-generated responses for safety, factual accuracy, policy compliance, and overall quality in a fully remote role. The position offers contract terms and the ability to contribute to cutting-edge AI safety benchmarking.

You will review content involving misinformation, political persuasion, self-harm, violence, cyber, biosecurity, and other sensitive domains; apply evaluation rubrics for RLHF and SFT; identify unsafe outputs; and provide