AI Response Reviewer — Contractor
Reviewed AI-generated responses and provided quality judgments using structured rubrics for reasoning, factual accuracy, and logical coherence. Applied consistent evaluation to multi-step reasoning and chain-of-thought validation across high-volume technical batches. Documented model errors, inconsistencies, and failure modes in structured written feedback to improve precision and reproducibility. • Evaluated reasoning quality, factual accuracy, and logical coherence. • Produced rubric-based preference data to support RLHF model improvement. • Validated multi-step reasoning and chain-of-thought consistency. • Wrote structured error analyses for precision and reproducibility.