This is for a medior solution, but it’s more intended as a data engineer’s “fizzbuzz” and then in the follow-up call I’ll ask them how they would distribute the workload over multiple clusters, make it idempotent, etc — that’s where the real technical evaluation happens.

"extract, transform, load" - it's an acronym often used to describe typical data wrangling from one source to another