Conceptual Map

The Four Layers of AI Systems

A conceptual map of the entire wiki - from raw AI capabilities down through engineering patterns, software foundations, and infrastructure.

1
AI Capabilities

What AI can do and how to harness it

The top of the stack is the AI itself - the models, techniques, and capabilities that make intelligent systems possible. This layer covers the core building blocks: language models, embeddings, agent architectures, evaluation, and the craft of prompt engineering.

2
Engineering Patterns

How to architect and wire AI into systems

Raw capabilities need structure. This layer covers the architectural patterns and engineering disciplines that turn AI model calls into reliable, maintainable production systems - RAG pipelines, orchestration, tool-calling, observability, and prompt lifecycle management.

3
Software Foundations

The engineering discipline beneath the AI

AI systems are software systems. Solid foundations in version control, testing strategy, CI/CD, API design, security, and architectural principles are what separate prototypes from production. This layer is the bedrock every AI engineer should have in place.

4
Infrastructure

The compute and data layer that runs it all

Every AI system ultimately runs on infrastructure - cloud services, containers, data pipelines, vector stores, and storage. This layer covers the operational reality of deploying and scaling AI: where data lives, how it moves, and what it costs to run.

Start exploring any layer

Each layer builds on the ones below it. Dive in where you are, or work top-to-bottom for the full picture.