AI Data Trainer & Annotation Contributor (Freelance / Contract)
Trained and annotated large volumes of supervised ML data for NLP and conversational AI, including text, audio transcripts, and structured inputs. Performed RLHF activities by ranking and rating model responses using the HHH (helpfulness, honesty, harmlessness) framework. Ensured outputs met instruction-following and safety requirements through continuous quality checks and policy compliance enforcement. • Labeled 15,000+ samples for supervised model training across NLP and conversational domains. • Rated/ranked responses for helpfulness, accuracy, harmlessness, and honesty under RLHF workflows. • Flagged edge cases, adversarial inputs, and policy-violating content to support red-teaming and safety alignment. • Achieved 97%+ inter-annotator agreement via consistent guidelines and quality assurance processes.