Ryan Knight

Ryan Knight is a Senior Partner Architect at Neo4j, building enterprise AI agents that use knowledge graphs to understand relationships across organizational data and retain context between interactions. His work with AWS, Databricks, Microsoft’s agent framework, and GitHub Copilot makes graph intelligence accessible within existing enterprise platforms.

Knight entered Java consulting at the Sun Java Center in 1999 and subsequently held architecture and advisory roles involving DataStax, Lightbend, Starbucks, and Capital One. By 2017, he was an enterprise architect at Starbucks. He later led Grand Cloud, a consultancy delivering cloud, data, and generative-AI systems for organizations ranging from startups to Fortune 500 companies.

At Neo4j, he works with cloud and data-platform partners on knowledge-graph-grounded enterprise agents. His projects address several specific architectural problems:

• Connected context beyond similarity search: His co-authored Microsoft Agent Framework integration combines vector retrieval with graph traversal, allowing agents to follow relationships among companies, products, geographic exposures, and risk factors.
• Persistent agent memory across sessions: The same integration retrieves relevant conversational history before model execution and stores messages, extracted entities, user preferences, and reasoning traces afterward, giving agents continuity across interactions.
• Graph intelligence inside the lakehouse: His Databricks fraud-investigation work makes graph-derived account scores and fraud-community signals available to Databricks Genie as queryable lakehouse dimensions.
• Agent-ready access to real graph schemas: A GitHub Copilot integration uses Neo4j’s Model Context Protocol tooling to inspect a database schema and generate Python code grounded in its actual nodes and relationships.

At AI Engineer World’s Fair 2026, Knight joined Zach Blumenfeld and Ben Squire for a hands-on lakehouse workshop on applying graph-shaped context to structured warehouse records and unstructured documents. His role centered on helping participants put those shared techniques into practice.