DriftLedger

DriftLedger is a web app (with optional CLI) that monitors data quality by focusing on what breaks in real life: silent drift. Instead of trying to be a full data catalog or another rules engine, it continuously profiles key tables/metrics, learns baseline distributions, and alerts when values, null rates, freshness, joins, or metric definitions shift beyond agreed thresholds. It plugs into common warehouses (Snowflake, BigQuery, Redshift) and BI outputs, then produces a simple “trust score” per dataset and per KPI. The MVP prioritizes fast setup: connect, pick critical tables/KPIs, and get drift alerts with plain-English explanations and suggested checks. It also keeps an audit trail of incidents, owners, and resolution notes so teams stop repeating the same firefights. This is an AI + traditional app: AI helps summarize anomalies and propose likely root causes, but deterministic checks remain the backbone.