Building and Deploying Multi-Agent Systems with ADK and Vertex AI

This hands-on workshop teaches participants how to build, orchestrate, and deploy multi-agent AI systems using Google’s Agent Development Kit (ADK) and Vertex AI Agent Engine. Attendees will learn how to create AI agents powered by Gemini models, build custom tools, and implement sequential, iterative, and parallel workflows. The workshop explores design pattern trade-offs, agent architecture decisions, callbacks and observability, memory management, and scalable deployment. It is structured around 10 progressive Jupyter notebooks and concludes with a full-stack capstone project deployed to Google Cloud using Agent Engine for agents and Cloud Run with a React and FastAPI stack for the frontend and backend.

Key takeaways • Build and orchestrate multi-agent AI systems with Google ADK • Create custom tools and coordinate agent teams powered by Gemini models • Choose between LLM-driven and workflow-driven design patterns (Sequential, Parallel, Iterative) • Navigate agent architecture trade-offs: Sub-agents vs. Agent as a Tool • Implement callbacks, observability, and guardrails for production readiness • Master session management, state, artifacts, and short-term vs. long-term memory • Deploy production-ready full-stack AI systems on Google Cloud (Agent Engine + Cloud Run)