LLM Safety Evaluator (Hebrew & English Required)

Project Overview / Instructions

Candidates must have a bachelor’s degree or equivalent professional experience in a relevant field, with near-native or native proficiency in Hebrew and at least C1 proficiency in English. Required qualifications include experience in trust and safety, content moderation, LLM red teaming, and strong familiarity with safety domains such as hate, harassment, violence, sexual content, self-harm, and misinformation. Applicants must demonstrate strong judgment, resilience to explicit content, and prior experience using AI tools such as Perplexity, Gemini, or ChatGPT. Project contributors will annotate, curate, and review model responses for safety, accuracy, and policy compliance. Work includes generating adversarial cases to test model robustness, scoring responses, documenting failure modes, and maintaining annotation standards. This work is critical to preventing unsafe outputs and involves collaboration with a global team to improve leading AI models.