At Freelance, I evaluate AI-generated outputs for factual accuracy, reasoning quality, instruction-following, and safety. I use Python, SQL, Excel, and Power BI to analyze datasets and visualize recurring model failure patterns.
At e2f, Inc., I reviewed and annotated AI training videos against detailed grading guidelines. I completed 60+ assessments daily while maintaining approximately 95% inter-annotator agreement, and documented ambiguous examples and edge cases.
I built an LLM Response Evaluation Framework with a 70-prompt, 210-evaluation dataset for comparative model assessment and failure analysis. Outside evaluation work, I practice photography and videography across street, wildlife, automotive, aviation, and landscape subjects, and publish visual work on TikTok.
