---
title: "AI Systems"
url: "/learn/ai-systems"
---

Everyone talks about the model. Almost nobody talks about the pipeline that
feeds it, the eval that catches when it drifts, or the failure modes that only
show up once real traffic hits it — and that gap is where most production AI
work actually happens.

This category is about that surrounding engineering: retrieval, evaluation, and
the discipline of building systems around a model rather than just calling one.

## Subcategories

- [Retrieval](/learn/ai-systems/retrieval.md): Retrieval is a search problem wearing an AI costume, and most of what makes a RAG pipeline good or bad happens before the model ever sees a token — how the corpus was chunked, what got indexed, and whether the right passage was even a candidate. These posts work through that layer directly, with evals rather than vibes.
- [Evaluation](/learn/ai-systems/evaluation.md): An eval is a test suite, and treating it as an afterthought produces the same thing skipping tests always produces — a system nobody can change with confidence. These posts treat evaluation as the spec an AI feature is built against, written before the prompt rather than bolted on once something has already shipped.