https://drive.google.com/file/d/1vFDpYJH0tHc2eDb6d-JGTWmo6Yb3PqF3/view?usp=share_link 

Traditional Retrieval-Augmented Generation (RAG) systems often fall short when queries require multi-step reasoning, relationship traversal, or contextual decision-making. In this hands-on workshop, participants will go beyond basic RAG by building an end-to-end, production-style knowledge system that combines hybrid retrieval, context graph construction, and multi-agent orchestration using AWS-native services.

In the first half of the session, attendees will construct a knowledge base pipeline using advanced chunking strategies, metadata enrichment, and hybrid search (vector + BM25). They will then extend this system by building a context graph powered by Amazon Neptune, enabling structured relationships across entities, documents, and systems.

In the second half, participants will design and implement a multi-agent system using Amazon Bedrock, AWS Lambda, and AWS Strands with MCP connection. These agents will collaborate to retrieve, traverse, and reason over both the knowledge base and graph, enabling more advanced query resolution and decision workflows.

The workshop will also demonstrate rapid development workflows using Strands and Kiro IDE to accelerate experimentation and iteration.

By the end of the session, attendees will have a working system capable of answering complex queries that require both semantic understanding and relational reasoning, demonstrating a practical path from traditional RAG systems to intelligent, agent-driven architectures.