technical gaurav tayade

At Outlier, I evaluate multi-turn LLM conversations for factuality, safety, context, and tone, while red-teaming models to uncover hallucinations, vulnerabilities, and edge cases. I also produce structured rationales under strict client rubrics and consistently achieve top-tier ratings.

Across independent AI data operations projects, I've built quality frameworks for text classification, search relevance, and multimodal annotation. I translate complex specifications into practical training resources, track quality metrics, and deliver precise video, image, and audio evaluations under tight deadlines.