Transitioning from data analyst to data engineer involves more than just learning new tools; it requires adopting an engineering mindset. The author details building a second ETL pipeline, an automated RSS ingestion system, to explore this shift. Unlike the first project, a GitHub ET L pipeline that evolved from a simple script to a scheduled task, this endeavor focused on orchestration and reliability.

The RSS pipeline, using Python, PostgreSQL, and Docker, aimed to understand how workflow orchestrators separate execution from logic and manage failures. The author intentionally kept the ETL logic simple to concentrate on architectural decisions. Key steps included making PostgreSQL inserts idempotent with `ON CONFLICT DO NOTHING` and containerizing the application with Docker, treating the container as the unit of execution for the orchestrator.

This project introduced the author to Kestra, a workflow orchestrator designed for data engineering. The experience shifted the focus from 'how to run a script' to questions about retry configurations, execution tracking, and separating application logic from infrastructure. The author concludes that building simple, focused projects that tackle specific engineering challenges is more effective for learning than trying to grasp numerous complex concepts simultaneously.