Documentation

Introduction

The AI Engineering framework organizes the practice of building, running, and governing AI systems into five domains. Together they form a value chain — from technical foundation, through the systems that put it to work, to the people who use it and the value it creates — with governance wrapping every stage.

Infrastructure & Architecture

Foundation & Build

Orchestration

System Integration & Workflow

Who Does This Work

Three roles get lumped together as “AI jobs,” but they build different things:

Role	What they build	Day-to-day focus
AI Researcher	New algorithms and model architectures	Novel training methods, papers, benchmark design
ML Engineer	Models trained from scratch on their own data	Feature pipelines, training runs, model serving
AI Engineer	Products built on models that already exist — GPT, Claude, Llama	APIs, retrieval, agents, evaluation, cost, safety

This framework is written for the third role. It treats the model as a component that already works, so the engineering problem is everything around it: what you feed it, what you let it do, who uses it, what it costs, and how you keep it in bounds.

Learning Progression

The five domains are organized by architecture, not by difficulty. If you are building the skill set rather than looking up a specific topic, work through it in this order instead:

    flowchart LR
    S1["1<br/>Python"] --> S2["2<br/>AI & ML Basics"]
    S2 --> S3["3<br/>Model APIs"]
    S3 --> S4["4<br/>RAG"]
    S4 --> S5["5<br/>Agents"]
    S5 --> S6["6<br/>Production Skills"]

    style S1 fill:#EFF6FF,stroke:#2563EB,color:#1E40AF
    style S2 fill:#EFF6FF,stroke:#2563EB,color:#1E40AF
    style S3 fill:#2563EB,stroke:#1D4ED8,color:#fff
    style S4 fill:#7C3AED,stroke:#6D28D9,color:#fff
    style S5 fill:#16A34A,stroke:#15803D,color:#fff
    style S6 fill:#EA580C,stroke:#C2410C,color:#fff

Stage	What you learn	Where this framework covers it
1. Python	Variables, control flow, functions, data structures, file I/O, working with libraries	Prerequisite — assumed by every code example on this site
2. AI & ML basics	How machine learning and neural networks work; what an LLM actually does (next-token prediction); prompting; why hallucinations happen	AI Model Fundamentals, Prompt & Context Design, AI Literacy Education
3. Model APIs	Talking to a model from code rather than a chat window; structured outputs; function calling	Agent Interface, Model Selection & Tuning
4. RAG	Chunking and embedding your own documents so the model can answer from information it was never trained on	RAG Pipeline, RAG 2.0, Vector DB Optimization
5. Agents	Goals, multi-step planning, calling tools and external APIs, carrying state across steps	Agent Interface, State Management, Workflow Automation
6. Production skills	Version control, cloud deployment, vector databases, evaluation and testing, security and privacy	AI-Native Design Framework, Compute Resource Management, AI Model Benchmarking, Frontier Benchmarks & Evaluation, Guardrails & Security

Stages 1–3 are foundations; the rest of this site starts where they leave off. For the developer track and practice projects that exercise these stages end to end, see AI Literacy Education.

Two competencies sit alongside this progression rather than inside it. Fine-tuning and quantization matter once prompting and retrieval have been exhausted — see Model Selection & Tuning. Serving and inference infrastructure matters as soon as you host a model yourself rather than calling an API — see Compute Resource Management.

Outside Reading

This framework covers the engineering practice, not the fundamentals underneath it. These open courses do, and map onto the stages above:

Stages	Resource
2	Microsoft — Generative AI for Beginners
2	DAIR.AI — Prompt Engineering Guide
3, 4	Awesome LLM Apps — reference implementations to read
5	Microsoft — AI Agents for Beginners
2, 6	Maxime Labonne — LLM Course — model internals, fine-tuning, quantization
2	Stanford CS229 — Transformers, In-Context Learning — the lecture behind the tokenization, attention, and decoding sections here