Sponsored by Soda | How Nubank Automates Data Quality Scores with Soda

Nubank, one of the world's largest digital banks, runs a massive Spark-based data platform. For years, every team built their own data quality validations. The result: duplication, gaps, and no shared definition of trust. This session shows how Nubank used Soda to standardize baseline data quality checks across all public P1 and P2 datasets. Redundant work disappeared, and every team got a common foundation. Once these checks went live in Soda Cloud, adoption took off: checks grew ~500%, monthly active users grew 48%, and teams started authoring event-driven custom checks and business rules at scale. On that foundation, Nubank built a Data Quality Score measuring the operational reliability of public P1 and P2 Spark jobs over a rolling 40-day window using weighted standard-check results. Data quality went from a per-team concern to an organization-wide KPI. Attendees leave with a blueprint for going from ad-hoc validations to a scored, scalable data quality program on Databricks.