Open Source Reporting Pipeline

You work as a data engineer at Spotify. The executive team requests a weekly analytics report summarizing user engagement and streaming trends, using data from internal databases and log files. Strict budget constraints prohibit the use of commercial or managed cloud services; you must rely solely on open-source technologies such as PostgreSQL, Apache Airflow, Apache Spark, and Metabase.

How would you architect this reporting pipeline from data ingestion to report delivery?