Freelance AI Trainer & RLHF Annotator
Conducted RLHF data annotation on AI training platforms, ranking and rating model responses for helpfulness, harmlessness, and factual accuracy. Evaluated prompt–response pairs specifically within HR, legal, and workplace compliance domains, providing calibrated preference feedback for model improvement. Authored adversarial and edge-case prompts to assess and expose model weaknesses in professional and compliance scenarios. • Reviewed over 500 examples for inter-annotator agreement and labeling consistency. • Flagged unsafe or biased model outputs, contributing to AI safety and content moderation goals. • Utilized annotation interfaces such as Surge, Scale AI, and Label Studio for quality assurance. • Specialized in tone, clarity, legal accuracy, and policy-focused instruction evaluation.