AI Safety LLM Evaluator (French/English, LLM Red Team Exp Required)

Project Overview / Instructions

Candidates must have a bachelor’s degree or higher in a relevant field, near-native or native French proficiency, and C1-level English. Required experience includes LLM red teaming, trust & safety, or content moderation, and significant familiarity with safety domains such as hate, harassment, sexual content, self-harm, and misinformation. Strong judgment, emotional resilience, and familiarity with AI tools like Perplexity, Gemini, or ChatGPT are essential. You will curate and annotate safety-focused training examples in English and French, stress-test and audit AI model outputs, and apply written safety policies to complex or ambiguous cases. This includes documenting model failure modes, suggesting decision rules, and helping ensure annotation consistency to support safer large language models.