As a freelance AI Evaluator and Multimodal Content Reviewer, I’ve evaluated 3,500+ AI-generated outputs across images, video, audio, and conversational text, achieving 96% rubric alignment accuracy.
I assess text-to-image and reference-to-image results, including instruction following and reference preservation, and perform pairwise preference ranking with structured justifications.
I’ve authored 800+ critiques of LLM and diffusion model responses, identifying hallucinations, factual errors, and safety violations. I’ve also engineered and stress-tested 600+ evaluation prompts across visual and conversational domains.
Earlier, I researched and edited long-form articles, audited user-generated content, and reviewed data entries for accuracy and guideline compliance. My projects include a 100-conversation multimodal evaluation set and a benchmark comparing DALL-E 3, Midjourney v6, and Stable Diffusion XL.

