Orchestrating Success

This title could be clearer and more informative.Try out Clickbait Shieldfor free (5 uses left this month).

Vinted solved the challenge of coordinating data pipelines across decentralized teams by building a DAG generator that automatically creates task-per-model Airflow pipelines from dbt manifests. They developed an Asset Registry to manage cross-domain dependencies using ExternalTaskSensor, enabling fine-grained task-level coordination without manual wiring. The system automatically handles late completions and timed-out sensors through completion callbacks, while providing CI/CD visibility into downstream dependencies. This approach standardizes pipeline creation, eliminates manual DAG authoring, and allows platform-wide changes like the Airflow 3 upgrade to be rolled out transparently.

11m read timeFrom vinted.engineering
Post cover image
Table of contents
The Dark Side of DecentralizationThe Rise of our DAG GeneratorDecentralized Domains, Centralized DependenciesTurning Decentralized Data Modelling into a BreezeWhy Everyone Needs a DAG GeneratorAppendix

Sort: