Hebrew Trust & Safety Data Trainer | $50/hr Remote

Hebrew Trust & Safety Data Trainer Work Snapshot

• Job Type: Contract
• Location: Remote
• Compensation: Up to $50 per hour
• Commitment: Flexible / project-based (hourly)
  
  

• Review and evaluate AI-generated safety-related content and reasoning for accuracy, policy compliance, clarity, and contextual safety
• Curate and label Trust & Safety training datasets in both Hebrew and English across areas such as hate speech, harassment, self-harm, violence, misinformation, malicious activity, and harmful content
• Perform red-teaming and adversarial testing to identify vulnerabilities, unsafe outputs, policy loopholes, and edge-case behaviors in AI systems
• Compare and rank AI-generated responses based on safety alignment, reasoning quality, risk severity, and adherence to moderation policies
• Document unsafe behaviors, escalation patterns, evasion attempts, and procedural risks with clear written rationales and structured evaluations
• Analyze and interpret nuanced Hebrew and English content while maintaining cultural, contextual, and linguistic accuracy
• Support AI model improvement through policy enforcement, annotation consistency, safety audits, documentation, and quality assurance workflows
• Flag ambiguous scenarios, recommend policy refinements, and contribute to maintaining high annotation standards across review teams
  
  

• Education: Bachelor s degree or higher in Communications, Linguistics, Psychology, Law/Policy, Security Studies, or a related field; equivalent professional experience also considered
• Native or near-native Hebrew proficiency with strong English communication skills (C1 or above) for policy interpretation and multilingual evaluation tasks
• Professional experience in Trust & Safety, content moderation, policy enforcement, investigations, compliance, risk operations, or safety evaluation workflows
• Strong understanding of safety domains including hate & harassment, self-harm, sexual content, violence, misinformation, malicious activity, and bias assessment
• Proven LLM red-teaming or adversarial testing experience with the ability to identify vulnerabilities and document mitigation strategies
• Ability to handle explicit, toxic, violent, or psychologically disturbing content while maintaining consistent judgment and professionalism
• Strong analytical writing skills with attention to detail, policy consistency, and structured decision-making across large-scale review workflows
• Familiarity with AI training, annotation, evaluation tools, and platforms such as ChatGPT, Gemini, Perplexity, or similar AI systems preferred