Data Annotator
DocumentText Generation
Evaluated LLM responses using rubrics (accuracy, relevance, clarity, safety) and wrote concise justifications. Performed ranking and comparison tasks to improve preference data. Flagged policy violations and low-quality outputs while maintaining consistent guideline adherence.
Evaluated LLM responses using rubrics (accuracy, relevance, clarity, safety) and wrote concise justifications. Performed ranking and comparison tasks to improve preference data. Flagged policy violations and low-quality outputs while maintaining consistent guideline adherence.
2025 - Present