Workshops

This workshop is a hands-on training program to develop serverless skills using AWS SAM and SAM CLI. It focuses on practical, code-centric learning with real-world scenarios, covering distributed event-driven architectures, messaging patterns, orchestration, and observability. Participants will use open-source tools, Powertools for AWS, and CI/CD. Available in Python, TypeScript, Java, or .NET.

A new version of this workshop has been published on Aug 6. Basic knowledge of AgentCore fundamentals is a prerequisite. Please complete the pre-work of pre-testing the workshop and identifying the right subset of samples for your audience. Use the materials in the workshop studio - videos, blogs and reading material to prepare before delivering or attending the workshop. Claude Code is configured with AWS documentation MCP servers as your learning partner.

Discover Kiro, your IDE build from the ground-up to seamlessly integrate AI agents into the development workflow. Explore how to use Kiro as your intelligent partner - helping to understand complex codebases, crystalize detailed requirements, and deliver quality solutions at unprecedented speed. You will iterate with Kiro to finalize your detailed design, before using Kiro to orchestrate AI agents and implement a fully working solution

Getting started on AgentCore: Let's build an end-to-end use case using Amazon Bedrock AgentCore.

Turn plain English into specs, working code, and tests — reviewing every step. Build a real app hands-on with Kiro, Amazon's AI-native dev agent, on your surface of choice: IDE, CLI, or IntelliJ. Go from chat-driven edits to full spec-driven development, then steer the agent with skills, MCP, custom agents, and hooks. No backend to deploy — it runs in your browser. ~4 hours, modules stand alone. For developers, from first-timers to automation power users.

Give your coding agent the tools, knowledge, and guardrails to build production-ready infrastructure on AWS.

In this hands-on workshop, you build, test, and optimize an agentic self-service experience in Amazon Connect Customer using the agentic CX designer (ACXD). You start in a prebuilt Amazon Connect environment, pick one of ten industry scenarios, and take a conversational agent from its first prompt to a tested, deployed experience on chat and voice. No coding is required.

Learn to track and optimize generative AI spend on Amazon Bedrock using multiple cost attribution methods: IAM principal attribution, inference profiles, projects, per-request metadata tagging, and IAM identity log attribution. Apply these across multi-tenant apps, Claude Code, and AgentCore agents. Explore LiteLLM for multi-provider tracking. Build dashboards with Cost Explorer, CUR 2.0, and CUDOS.

Build a multi-agent AI system on Amazon EKS Auto Mode that serves every model on CPU: a fine-tuned intent router and embedding model (ONNX, trained on CPU), a cross-encoder re-ranker, and a quantized generative analyst on llama.cpp grounded by RAG from OpenSearch. Amazon Bedrock generates synthetic training data, judges answers with a faithfulness gate, and handles edge cases small models cannot. A control test puts a frontier LLM in the same seat so you know if the small model is good enough.

The RDS PostgreSQL workshop provides customers with hands-on experience using Amazon RDS for PostgreSQL. It uses a modular content format including foundational, intermediate, and advanced labs covering a wide range of RDS features and PostgreSQL topics.