PipeSleuth

PipeSleuth is a web app that continuously inspects, profiles, and stress-tests data pipelines across warehouses, lakes, and streaming systems. Instead of waiting for angry stakeholders to report broken dashboards, it uses AI to learn normal data patterns and pipeline behavior, then flags anomalies, schema drifts, and silent data corruption before they hit production. It connects to tools like Airflow, dbt, Snowflake, BigQuery, Kafka, and Fivetran, building a dependency graph and impact map so teams see exactly which tables, reports, and services are affected by an upstream issue. The app focuses on ruthless practicality: clear incident timelines, root-cause hints, and a ranked list of likely failure points, not fluffy AI magic. It’s designed for teams already drowning in DAGs, half-documented jobs, and brittle handoffs between data engineers and analysts. PipeSleuth won’t magically fix bad architecture, but it will make it brutally obvious where things are breaking, how often, and what to fix first. This is an AI-first web app, with a small companion CLI/agent for on-prem environments.