The AI Architecture Bootcamp is a free, hands-on course for software engineers who want to build AI systems that hold up in production, mostly on AWS. It is organized as 10 modules, each with a project. Every module is one guide with tested code, diagrams, troubleshooting and an FAQ, and every claim is checked against the official documentation.

How to use this bootcamp

  1. Work through the modules in order; each builds on the previous one.
  2. Run the code as you read. Examples run locally; AWS calls are marked, and their request shapes are checked against the current SDK.
  3. Finish each module with its project, then use the troubleshooting table when something breaks in your own system.

Available modules

Module 1: LLM Foundations for Coding Systems

Days 1 to 10. A documentation assistant that answers from your runbooks, cites what it used, refuses without evidence, and has an eval gate.

Covers: Tokens and context budgets, the KV cache, embeddings, vector indexes, a debuggable RAG pipeline, chunking, prompt contracts, structured outputs, evals.

Prerequisites: Python and basic AWS familiarity. Reading time: about 30 minutes, plus time to run the code.

Module 2: Agents, Tools and Production Guardrails

Days 11 to 20. An agent loop with typed tools, budgets, memory tiers, guardrails and approval gates, ending in a diff-scoped PR review bot.

Covers: Tool contracts, planning patterns, memory, Bedrock Guardrails, token cost curves, honest streaming, human approvals, safe logging, online evaluation.

Prerequisites: Module 1. Reading time: about 32 minutes, plus time to run the code.

Module 3: Production RAG for Code

Days 21 to 30. Multi-tenant code search on Bedrock Knowledge Bases with hybrid retrieval, reranking, tenant isolation and checked citations.

Covers: Hybrid search with RRF, rerankers, metadata filters, safe query rewriting, call-graph expansion, citations, re-indexing on push, multimodal inputs, a failure taxonomy.

Prerequisites: Modules 1 and 2. Reading time: about 32 minutes, plus time to run the code.

Coming soon

These modules are in preparation and will be linked here when they are published.

  • Module 4: Serving AI on AWS: Lambda, Bedrock, Step Functions, EventBridge (Days 31 to 40)
  • Module 5: Fine-Tuning and Model Customization: When It Beats Prompting (Days 41 to 50)
  • Module 6: Multi-Agent Systems Without Chaos (Days 51 to 60)
  • Module 7: Security for AI Coding Agents (Days 61 to 70)
  • Module 8: Cost, Latency and Scale for AI Workloads (Days 71 to 80)
  • Module 9: Memory, Multimodal and Domain Copilots (Days 81 to 90)
  • Module 10: AI Architecture Leadership and Capstone (Days 91 to 100)

Deep dives

Standalone guides that go further on topics the modules touch:

Lessons for upcoming modules

Until each module above is published, its individual day lessons stay available here.

Module 4: Serving AI on AWS: Lambda, Bedrock, Step Functions, EventBridge

Module 5: Fine-Tuning and Model Customization: When It Beats Prompting

Module 6: Multi-Agent Systems Without Chaos

Module 7: Security for AI Coding Agents

Module 8: Cost, Latency and Scale for AI Workloads

Module 9: Memory, Multimodal and Domain Copilots

Module 10: AI Architecture Leadership and Capstone

Watch: 100 Days of AI on YouTube

Short videos from CheatCoders, one AI engineering topic per day.

Open the playlist on YouTube · CheatCoders on YouTube

Last updated October 10, 2026