Platform Engineering

The internal developer platform is the most reliably over-built system in modern engineering organisations. It is funded on the promise of leverage, staffed by infrastructure engineers who have never run a product, measured on delivery of features nobody requested, and abandoned when adoption stalls at thirty percent of eligible teams. The technology is rarely the problem.

This section treats platform work as product work, because the constraints are identical: you have optional users, a competitor (the status quo of raw cloud access), and a discovery problem you cannot solve by reading Slack. Covered here — the abstraction boundaries that hold and the ones that leak, why the golden path becomes a cage, how multi-cloud portability gets sold and what it actually costs in practice, and the specific anti-patterns that show up in nearly every failed platform: the mandatory abstraction, the YAML-generating YAML, the ticket queue wearing a self-service costume. Positions are stated plainly. Some of them are unpopular.