Multi-Agent Evaluation Workshop (In Person, Munich)

​Learn to build and evaluate multi-agent AI systems. Using Strands SDK, Amazon Bedrock, and AgentCore, you'll construct a four-agent graph workflow (document analysis, policy retrieval, risk inspection, and claim summarization), then measure its performance through model, operational monitoring and quality evaluation.

​📌 Agenda Highlights

• ​Multi-agent orchestration with graph-based routing

• ​CloudWatch dashboards for GenAI metrics (tokens, latency, cost)

• ​Programmatic accuracy testing (Precision, Recall, F1)

• ​LLM-as-Judge evaluation(model evaluation) with 10+built-in and custom evaluators

​🎯 This is intended for intermediate Python developers/engineers with AWS experience, interested in building and evaluating AI agent systems. Familiarity with REST APIs and JSON required; ML metrics knowledge helpful but not mandatory.