From click‑ops and custom apps to code
Managing the dbt at scale isn't about creating projects; it's about building a data foundation all teams can trust.
Our organization supports ~3,000 dbt platform projects across 6 accounts and two VPCs: one for DEV, one for PROD. Until recently, deployments depended on a custom-built promotion system we'd cobbled together to work around restricted UI access in PROD. Every new feature dbt Labs shipped meant we had to rebuild internal components before our teams could use it.
This session walks through how we replaced that entire system with a full Terraform implementation built on the official dbt platform provider. We'll cover our design decisions, module patterns, drift-elimination strategies, and the onboarding model that lets teams own their infrastructure through GitOps with consistency, auditability, and self-service built in.
Check out more sessions
- Training add-on
Mastering data quality with dbt
Carol Ohms / dbt LabsStephen Thibeault / dbt LabsView session - Breakout session
YAML doesn't know why: building the business context your agents are missing
Pedro Heyerdahl / Kilo CodeView session - Breakout session
One engine to rule them all: why Obie left the shire
Tyson Doberneck / Obie InsuranceMatt Karan / Obie InsuranceView session
