Architecture & Design Principles — Platform Engineering at Scale

A field-tested distillation of how to build a developer platform that serves ~1000 services across ~13 regions on ~2000 long-lived proxies, derived from an 8-year platform-engineering retrospective (Atlassian, 2017-2025).

The talk is structured as a chronological build-up; this skill re-organises it as a principle map plus deep-dive references.

Why this skill exists

Platform-engineering content online tends toward two failure modes:

1. Vendor demos that show the happy path of a tool without the failure modes that emerged at year three.
2. Conference talks that describe a finished system without the sequence of decisions that produced it.