AI Trainer and Legal Domain Expert (Freelance/Project-based)
As an AI Trainer and legal domain expert, I conducted reinforcement learning from human feedback (RLHF) and preference ranking tasks on legal document outputs generated by LLMs. My responsibilities included evaluating AI-generated legal documents for factual accuracy, legal coherence, and proper statutory citations. I also assessed jurisdictional correctness and reviewed outputs for linguistic fluency in Mandarin and English. • Reviewed and rated legal contracts, briefs, and memos for AI model helpfulness, harmlessness, and honesty. • Identified nuanced legal reasoning errors and flagged harmful or risky responses during red teaming evaluations. • Applied legal expertise to QA tasks, ensuring culturally and contextually appropriate Chinese-language model outputs. • Used internal/proprietary tools for annotation, providing structured evaluation and feedback on AI outputs.