IA na Prática: BigQuery, RAG e Multi-Agentes no GCP

🚀 Another GDG Talks is coming! Get ready for a night of real applied AI: three technical talks on how to build, architect, and deploy AI systems on Google Cloud Platform. ☁️ GCP Data Cloud — Unified AI and Analytics Platform With Josias Leonardo Junior BigQuery and Google Cloud components are not isolated tools — they are an integrated ecosystem to accelerate the development and use of AI at scale. In this talk, you will understand how the platform positions itself as a secure and unified foundation for teams wanting to go from raw data to production AI: - The role of BigQuery in the GCP data ecosystem - How the platform components integrate - How to accelerate AI development securely and at scale 🔍 RAG in SQL: AI architecture on GCP With Leonardo Henrique Oliveira Pena RAG has six pieces: extraction, chunking, embedding, vector database, orchestrator, and service. The real question is where each piece should reside. Leonardo builds an entire RAG within BigQuery, in pure SQL, without new infrastructure — and then honestly shows where this architecture falls short: - The six pieces of a RAG and where each can run - Why build everything within BigQuery, in SQL - The real limits: latency, cost per row, and state - The boundary between a query and an agent 🤖 Taming Multi-Agents in Go with Phoenix With Guilherme Henrique da Silva Creating an agent is the easy part. The real challenge is deploying multi-agent systems in production with security and observability. Guilherme shows how to orchestrate agents with Golang and monitor them with Phoenix: - Orchestrating multi-agent systems with Go - Capturing data generated by agents in production - Versioning prompts and managing experiments - Structuring robust datasets without compromising performance It will be a night of real architecture, difficult decisions, and AI that goes into production.