Cricket Subject Matter Expert (AI Model Evaluation) - Mercor
Evaluated AI-generated responses for factual accuracy, statistical integrity, reasoning quality, and narrative relevance during live international cricket matches. Documented hallucinations, contextual gaps, and tactical misunderstandings by verifying outputs against player histories and live expert analysis. Designed cricket-specific prompt and evaluation queries grounded in match context and performance data to test model recall and understanding. • Read prompts and ranked multiple AI responses in real time based on accuracy and helpfulness • Flagged and recorded incorrect statistics, logical inconsistencies, and fabricated expert opinions • Built search prompts and evaluation questions using player profiles and match situations • Produced structured written justifications and feedback to improve response reliability