Data quality in production
Data quality is not a one-time cleanup. It is tests and monitors that run every time data moves.
Quality is something you enforce, not a state you reach once. Data drifts: a source changes a format, a job half-fails, a duplicate slips in. Without checks, the first person to notice is someone reading a wrong number in a report.
So we put tests in the pipeline itself. Row counts, types, ranges, uniqueness, and freshness are checked on every run, and a run that fails a critical check stops before bad data lands. Softer checks raise an alert instead of blocking.
In production we watch the same signals continuously, with clear thresholds and owners. When something drifts, the alert points at the exact table and check, so the fix is quick and the trust holds.