Monorepo surgery, CI discipline, semantic layers, and cost levers — Discord's petabyte dbt to Datadog's 5,000-person self-serve.
Past a few thousand models, dbt stops being a tool and becomes a platform you engineer: Discord's petabyte-scale surgery on dbt internals, Checkout.com's 27-project CI library, the Airbnb/DoorDash/Netflix semantic-layer canon, and the SQLMesh-era platform decisions.
The platform memo: decompose what managed products sell, judge frameworks on state and validation economics, extend instead of fork, and design the self-serve operating model.
The renewal quote lands and it's triple last year's — the vendor moved from seats to consumption pricing, and your usage grew into it. The team meeting produces four camps in fifteen minutes: rebuild it all on CI runners ("it's just a scheduler"), migrate to the new framework everyone's blogging about, fork the tool and take control, and do-nothing-renew-and-grumble. Every camp has a blog post proving its case. None has decomposed what you're actually paying for, what each exit costs to *operate*, or who maintains any of it in year three. That memo is this week.
The full week 4 brief is part of LeetData Pro.
