AI response evaluation and rubric-based scoring
Evaluated AI-generated prompt and response pairs for quality and safety using structured rubrics. Assessed outputs for accuracy, clarity, helpfulness, tone, and adherence to user instructions, and provided scored results with written rationale. Identified issues such as hallucinations, vague wording, weak logic, unsupported claims, and missed user intent, then delivered actionable improvement feedback. • Prompt and instruction analysis • Rubric-based scoring and quality review • Fact-checking and research to verify claims • Clear written feedback with tone and audience adaptation