This is the best advice I've seen here. The purpose of data engineering is to support a much larger effort related to scientific discovery or increasing bottom line revenue. Don't start with the language, tooling, libs and architecture, start with a purpose and goal. Most goals relate to mimicking a portion of human cognition in terms of pattern matching.

If you're building a house you don't start by picking out hammers and saws.

This takes being passionate about your data and its pre and postprocessing as a data engineer.