Backend Patterns, Cloud Deployment & Live Demo

A practical, end-to-end session on taking AI from prototype to production. We’ll cover proven backend patterns for integrating AI into real systems, strategies for deploying on the cloud, and best practices for scalability, monitoring, and reliability—ending with a live demo of a production-ready AI service in action.

Architecting Multimodal Graph RAG with ADK & Memory Bank

Most AI agents fail in production because they lack context and continuity. When the user says, "Fix that diagram I showed you last week," standard RAG breaks. This session dives into a production-grade stack on Google Cloud to solve the three pillars of agency: Perception (Gemini), Structure (Spanner Graph), and Identity (Memory Bank). We will walk through building an agent using the ADK that performs multi-hop reasoning across unstructured visual data and structured graph nodes. You'll learn how to implement an evolving Memory Bank that de-duplicates user insights in real-time, turning stateless chat into a stateful, intelligent partnership.