Workbook - AI Engineering Bootcamp

• Week 1 Objectives: Introduction to AI Engineering
• What is AI Engineering?
• How does AI Engineering differ from other technical roles?
• The Modern AI Stack
• Large Language Models (LLMs)
• APIs as the Foundation of Modern AI
• How AI Systems Are Built Today
• Week 1: Review and Delivery Session: Your first day as an AI Engineer, live

• Week 2 Objectives: Python Applied to AI
• Python as a Tool for AI Engineers
• HTTP Fundamentals for AI Systems
• JSON in AI Applications
• Environment Variables and Configuration
• Options for Accessing LLMs: Local vs Cloud
• Consuming AI APIs with Python
• Understanding AI API Responses
• Structured Outputs
• Live Review and Deliverable — Week 2: Complete AI Chat Script

• Week 3 Objectives: LangChain
• The Problem: Why We Need LangChain
• What is LangChain?
• Prompts as Objects
• Chains: Connecting Steps
• Intro to Memory
• Thinking in Applications
• Weekly Deliverable
• Week 03 - LangChain: From Raw Code to Abstractions

• Weekly Objectives
• What is Prompt Engineering?
• Prompts as Behavior Contracts
• Zero-Shot Prompting
• Few-Shot Prompting
• Chain-of-Thought (CoT) Prompting
• Evaluating Prompts
• Prompts in Real Systems
• From prompt to context engineering
• Week Deliverable
• Week 04 — Prompt Engineering

• Goals for the week
• Capsule description
• What is FastAPI?
• Your First Endpoint with FastAPI
• Pydantic: Data Contracts
• Error Handling and Validation
• Automatic Documentation with Swagger
• Week 5 Delivery
• Week 5 — FastAPI: From Scripts to Services

• Week 6 Objectives: FastAPI + LangChain Integration
• AI API Architecture
• Integrating LLMChain into Endpoints
• State Management: Memory in APIs
• Async Patterns for LLMs
• Testing AI Endpoints
• Limitations of the Current System
• Week 6 Deliverable: AI Technical Assistant API
• Week 6 — FastAPI + LangChain: From Simulation to Real AI

• Week 7 Objectives: Introduction to RAG
• The Static Knowledge Problem in LLMs
• What is Retrieval-Augmented Generation (RAG)?
• Embeddings: The Foundations You Need
• Chunking: Splitting Information
• Your First End-to-End RAG Flow
• The Limitations of Basic RAG
• Week 7 Deliverable: Document Q&A Script
• Week 7 — RAG: From Static Knowledge to Real Documents

• Week 8 Objectives: Vector Databases
• Why We Need Vector Databases
• What Is a Vector Database?
• Introduction to ChromaDB
• Indexing Documents with ChromaDB
• Practical Semantic Search
• Integrating ChromaDB with FastAPI
• Week 8 Deliverable: Q&A API with ChromaDB
• Week 8 — Vector Databases: From Prototype to Production with ChromaDB

• Week 9 Objectives: Intermediate RAG
• Common Problems of Basic RAG
• Advanced Chunking Strategies
• Embedding Improvements (Practical)
• Structured RAG Pipelines
• Context and Length Control
• Evaluating Document Relevance
• Reranking, contextual retrieval, and the leap to agentic RAG
• Week Deliverable: Improved RAG with Advanced Techniques
• Week 9 — Intermediate RAG: Quality and Cost Optimization

• Week 10 — Agents I: from a manual loop to a stateful agent
• What is Function Calling / Tool Use?
• Why Function Calling is Critical
• Function Calling with OpenAI
• Building Safe Tools for your Agent
• The Agent Loop: Orchestrating Tools
• From the manual loop to StateGraph
• Tools, ToolNode and the ReAct cycle
• `create_react_agent` and structured outputs
• Persistence, memory, streaming and human approval
• Mini-project: your research agent
• Week 10 — Agents I: build your stateful agent live (LangGraph)

• Week 11 — Agents II: from an agent to a system
• Subgraphs and orchestration
• Multi-agent with a supervisor
• Direct handoffs and swarm
• Protocols: MCP and A2A
• Long-term memory by tiers
• Agent security and intro to evaluation
• Mini-project: your multi-agent system
• Week 11 — Agents II: build your multi-agent system live

• Week 12 Objectives: Evaluation & Observability of AI Systems
• Why Evaluation is Critical in AI Engineering
• Metrics for RAG: Relevance, Groundedness, Completeness
• Golden Datasets and Test Cases
• LLM-as-a-Judge: Practical Evaluation
• Benchmarking Awareness: MT-Bench, HELM, and Others
• Testing Agents and Tools
• Observability and Trajectory Evaluation
• Week Deliverable: Evaluate and Improve Your Agent
• Week 12 — Evaluation & Testing: A Professional Framework for AI Systems

• Phase 2 Project Goals
• Define Scope and Architecture
• Setup: RAG + ChromaDB + FastAPI
• Implement the Retrieval Pipeline
• Integrate Tool Use and Function Calling
• Apply the Evaluation Framework
• First Functional Version
• Checkpoint: Mid-Project Review
• Professional FastAPI Structure
• Week 13 — Build: Production-ready Phase 2 Project

• Refining the System Based on Metrics
• Optimizing Performance (Latency, Cost)
• Improving Evaluation and Coverage
• Complete Technical Documentation
• Preparing the Project Presentation
• Final Demos and Presentations
• Peer Review and Feedback
• Retrospective and Phase 2 Closing
• Week 14 — Polish: Data-driven optimization + Presentation

• Weekly Objectives
• Why Docker for AI Systems
• Docker Fundamentals
• Create Dockerfiles for AI Apps
• Docker Images and Containers
• Docker Compose for Multi-Container
• Environment Configuration
• Deliverable: Dockerized Phase 2 App
• Week 15 — Dockerize the Café Aurora rag-api

• Week Objectives
• Introduction to LocalStack
• Lambda-style Execution
• API Gateway-style Routing
• S3-style Object Storage
• Service Orchestration
• Deploy AI App on LocalStack
• LocalStack vs. Real AWS: Differences
• Delivery: App Deployed on LocalStack
• Week 16 — LocalStack: serverless patterns locally

• Week 17 Objectives: Production Deployment
• From Local to Cloud: The Big Leap
• Migrate the Vector DB: ChromaDB → Pinecone
• Migrate the LLM: Local → OpenRouter
• Deploy the Backend: Render.com
• Basic Monitoring in Production
• Testing and Troubleshooting in the Cloud
• Week Delivery: Complete Deployed App
• Week 17 — Production Deployment: Render + Pinecone + OpenRouter

• Goals for the Week
• Monitoring on Render Cloud
• Dashboards and Basic Metrics
• Why AI Systems Are Expensive
• Understanding LLM Usage and Cost Drivers
• Reducing Unnecessary LLM Calls
• Caching Strategies (Redis-style)
• Request Optimization
• Cost-Aware Design
• Delivery: Optimized System
• Week 18 — Monitoring & Cost Optimization

• Week Objectives
• Security Threats in AI Systems
• Secret Management
• Environment Isolation
• Access Control Concepts
• Prompt Injection Protection
• Secure API Design
• Agentic Security: OWASP Agentic and Red-Teaming
• Delivery: Secured System
• Week 19 — Security & Hardening

• Week Objectives
• Why Integrations Matter
• Working with Third-Party APIs
• Authentication and Tokens
• Webhooks and Event-Driven Flows
• Integration with Slack/Notion/Discord
• Designing Useful Integrations
• MCP and A2A: Standard Integrations for Agents
• Weekly Submission
• Week 20 — Slack bot: an AI agent in a real workspace

• Week Objectives
• Architectural Thinking for AI Systems
• Modular System Design
• Scaling Strategies
• Reliability & Fault Tolerance
• Trade-offs in AI Architecture
• Case Studies of Real Systems
• Weekly Deliverable
• Week 21 — System Design & Scaling

• Final Project Objectives
• Integrate All the Components
• Production-Ready Checklist
• Development and Integration
• Testing and Refinement
• Complete Documentation
• Final Presentations
• Bootcamp Wrap-up
• Week 22 — Capstone: Production-ready AI Engineer

For as long as your bootcamp enrollment is active. This guide is the cohort's material: it opens when you enroll and isn't sold separately.

No. Modules run from easier to harder, but you can jump to the one you need. Progress is saved per lesson.

Whatever is needed is listed under “What you need to bring”, above. If nothing is listed there, you can start from zero.

In your cohort's channel, with your instructor and the rest of the group. The bootcamp's live sessions are the other place to ask.

It's the bootcamp's material, not a replacement for it. The bootcamp adds dates, a cohort, reviewed work and support; this guide is what you study inside it.