A growth team's curated marts have no formal quality program

A medium Pipeline Design interview practice problem on DataDriven. Write and execute real pipeline design code with instant grading.

Domain

Pipeline Design

Difficulty

medium

Problem

A growth team's curated marts have no formal quality program. Each pipeline has ad-hoc checks (or none); thresholds are hard-coded; the same producer drifts every quarter; orphan keys silently understate revenue; nobody can answer 'is the data good' without a Slack thread. Build the quality program. Add schema validation at the source-to-raw boundary, distributional checks at the curated layer, referential-integrity checks between fct and dim, environment-aware thresholds, and the five-pillar suite at the serving layer. Block-class fires route to a paging destination; warn-class fires route to a non-paging notification channel. The dashboards keep reading from their existing Snowflake marts.

Practice This Problem

Solve this Pipeline Design problem with real code execution. DataDriven runs your solution and grades it automatically.