About

This is a reading room, not a dashboard. It distills engineering writing from about 30 company blogs into a cross-linked archive of concepts, patterns, and named systems.

How pages get here

An ingestion pipeline polls RSS feeds and HTML archives (for blogs without RSS) every few hours, downloading article markdown into raw/<company>/. Deduplication and title-normalization happen on the way in.

A pool of curator agents (running the Kiro runtime) then picks up each raw article and writes four things:

1. A source page under /sources/ — the article's distilled take, linking back to every concept, pattern, and system it references.
2. New or updated concept pages under /concepts/ — discrete ideas extracted from the article (e.g. eventual consistency, backpressure, blast radius).
3. New or updated pattern pages under /patterns/ — repeatable solutions (e.g. circuit breaker, saga, outbox).
4. New or updated system pages under /systems/ — named things that exist in production (e.g. Dynamo, Kafka, Spanner).

Curators are encouraged to cross-link liberally via Obsidian-style [[wiki-links]]. A post-processing pass resolves links, regenerates category indexes, and rebuilds the static site.

How to read it

Start from a concept. If you want to understand something, land on /concepts/<thing>/ — you'll get a focused page and a list of sources where the concept shows up in practice.

Start from a source. If you already know what article you want, browse /sources/ chronologically or hit Cmd-K and search. Source pages include the original URL.

Start with the most-cited. Each category index (/concepts/, /patterns/, /systems/) leads with the most-cited entries — the ones the corpus agrees matter.

What it isn't

• Not a tutorial. Each page assumes you know roughly what a distributed system is. Start with Designing Data-Intensive Applications if you don't.
• Not a replacement for primary sources. The whole point is to send you to the original engineering blog post. The wiki is a map.
• Not a news feed. Articles are ingested when they're interesting, not when they're published.

For agents

Every page on this site is also available as a plain-text catalog at /llms.txt for LLM consumption. The format follows the llms.txt convention.

Colophon

• Built with mkdocs-material, heavily customized.
• Typography: Fraunces (display, with its expressive italic) + Inter Tight (body) + JetBrains Mono (code).
• Color: OKLCH oxide-red palette. No indigo, no gradient text, no glassmorphism.
• Design direction follows the impeccable framework.