AI Response Evaluator at Tellus International AI
TextEvaluated over 1,000 AI-generated responses for clarity and relevance using detailed evaluation guidelines to ensure consistent, unbiased judgments. Produced structured feedback that improved the clarity of evaluations, supporting better training signals for AI model learning and raising performance by 25%. Assessed reasoning gaps and subtle nuances in responses, contributing to algorithm refinement and reducing response errors by 15%. • Assessed responses against predefined rubrics for clarity, relevance, and correctness • Provided written, evidence-based feedback aligned to evaluation criteria • Identified nuanced issues that could lead to errors or misinterpretations • Collaborated with a research team to analyze trends and inform training improvements