AI Data Trainer & Annotator (Freelance)
Freelance AI data trainer and annotator evaluating AI-generated text for accuracy, coherence, instruction-following, and safety compliance across multiple task types. Produced preference and quality signals by ranking and rating AI response pairs to support RLHF-style model alignment and fine-tuning. Documented detailed rationales while flagging failure modes such as hallucinations, factual errors, harmful outputs, and tone inconsistencies. • Evaluated 200+ hours of AI-generated outputs across summarization, Q&A, creative writing, and code explanation • Wrote and refined training prompts to probe capabilities and expose model failure modes • Ranked response pairs to indicate quality preferences for RLHF pipelines • Maintained 98%+ annotation accuracy and consistency under tight deadlines