Navin Sharma

Building reliable AI agents on Databricks requires more than pipelines and models — it demands a semantic foundation that encodes business meaning and context. In this session, we’ll show how to implement an ontology-powered semantic layer that integrates with Delta Lake and Unity Catalog to ground LLM-based agents. You’ll learn how to bootstrap ontologies, enable real-time semantic queries and go beyond traditional RAG approaches. We’ll walk through patterns for evaluating agent performance against natural language queries across use cases like supply chain, drug discovery and compliance. Leave with a practical architecture for deploying explainable, production-ready agentic AI on Databricks.