Analyst - Innodata Inc. (LLM evaluation and annotation)
TextEvaluated and annotated AI-generated responses for LLM training to assess response quality against relevance, accuracy, instruction-following, completeness, and user-intent alignment. Created and applied evaluation rubrics and annotation guidelines to standardize quality assessment for AI response moderators and to categorize errors for model refinement. Performed detailed analysis of model outputs to identify hallucinations, inconsistencies, reasoning gaps, and failures, providing structured feedback for behavior improvement. • Assessed conversational AI outputs and error types for downstream training improvements • Supported human feedback and manual data review within AI training workflows • Conducted quality assurance checks to improve dataset reliability • Provided structured feedback to improve LLM response quality