Flowsense Engine - Temporal drift and anomaly detection for Apache Airflow

Airflow provides rich execution metadata, but identifying behavioral drift across historical runs still requires manual analysis.

FlowSense is designed to answer questions such as:
- Which task started behaving differently?
- How large is the deviation from its historical baseline?
- Is the anomaly isolated or affecting downstream tasks?
-Where is the most likely origin of the slowdown?

I built it to help teams understand where slowdowns begin and how they spread.

I’d love to connect with people who care about reliable data pipelines and open-source tooling. Contributions, ideas, and feedback are always welcome.