Dungeons and Databases Build NPC agents to work with data in DocumentDB and Postgres (Sponsor: Microsoft)

When working with databases, developers face a labyrinth of choices: relational, NoSQL, vector, and graph. This session guides you through the latest advances in PostgreSQL on Azure, Microsoft's open-source DocumentDB project (a MongoDB-compatible NoSQL engine built on PostgreSQL), and powerful database extensions in Visual Studio Code. You'll see how managed PostgreSQL on Azure unlocks vector search (PGVector, DiskANN), graph queries (Apache AGE), and semantic search—and how VS Code's PostgreSQL extension lets you query, visualize schemas, and even chat with your database using GitHub Copilot. Throughout the talk, we'll build Python-powered data agents that work with both PostgreSQL and DocumentDB. You'll learn to combine vector and keyword search for hybrid retrieval, use graph relationships to surface insights, wire up DocumentDB for flexible session storage, and give your agents persistent memory—all from your favorite editor VS Code.