AI Bootcamp

From plain LLMs to RAG, agents, the AWS agentic stack, MCP and A2A, the Strands agent loop, and a hands-on Hello World agent on Amazon Bedrock.

Turn Python functions into agent tools with @tool, combine multiple tools, use the prebuilt Strands toolbox and use_aws, and build custom, class-based, and async tools.

Short-term and long-term agent memory, memory on AWS, a personalised Strands agent with Mem0, context management, and Amazon Bedrock AgentCore Memory with a LangGraph checkpointer.

Multi-agent patterns with Strands Agents on Amazon Bedrock: supervisor-worker, agent-as-tool, sequential, parallel and collaborative communication, and swarm and graph orchestration.

Model Context Protocol (MCP) for connecting agents to tools, resources, and prompts, and the Agent-to-Agent (A2A) protocol for connecting agents to each other with FastMCP and Bedrock labs.

Taking Strands agents from laptop to production on AWS Lambda, Amazon ECS on Fargate, and Amazon Bedrock AgentCore (Runtime, Gateway, Identity), plus enterprise integration patterns.

Evaluate agents with AgentCore Evaluations and LLM-as-a-judge, trace with OpenTelemetry into Jaeger and Langfuse, and govern with Bedrock Guardrails and Cedar.

Four real-world AI engineering project walkthroughs with 24 follow-up questions and 50 practice questions covering architecture trade-offs, RAG, agents, memory, and senior interview prep.

From a 60-line demo to a production-ready RAG system: document ingestion, PDF text extraction and OCR, chunking strategies, embeddings, hybrid retrieval, reranking, and grounded evaluation.

Build an AI team that answers questions from your PDF, routes to the internet when answers are missing, checks hallucination graders, and corrects its own work with line-by-line walkthroughs.

Design and build multi-agent deep research systems: query planning, parallel web exploration, source credibility checking, cross-referencing conflicting facts, and synthesising cited reports.

Finance domain concepts and hands-on agent labs: earnings-call JSON summariser, live price agent, financial ratio retriever, rebalancing assistant with memory, and Buy/Hold/Sell pipeline.

A five-agent Net Promoter Score (NPS) analyst for banking: promoters, passives and detractors, revenue at risk, root-cause theme clustering, and leadership action briefs.

End-to-end capstone system: automated vendor risk assessment combining a production RAG policy library, multi-agent router with graders, real-time web deep research, and human-in-the-loop review.