Eval++: Why LLM Evaluation Alone Isn't Enough

In this 25-minute conference talk from MLOps World, Chinar Movsisyan, Founder & CEO of Feedback Intelligence, introduces "Eval++," a new approach to evaluating large language models that goes beyond traditional metrics. Learn why conventional LLM evaluation methods that rely solely on internal data science metrics and offline benchmarks fall short in ensuring products meet real user needs. Discover how to place users at the center of development by transforming implicit feedback into actionable insights, understand the importance of closing the feedback loop to move from proof of concept to return on investment, and explore how both enterprises and startups are implementing Feedback Intelligence to build LLM-powered products that users genuinely trust and appreciate. Movsisyan brings her extensive experience building AI solutions in mission-critical applications including drones, satellites, and healthcare to this insightful presentation.