>working on problems that "don't matter"

Usually low level engineers don't have a say on what projects they get to work. They show impact by completing whatever projects were handed to them and hope/pray those projects take off and become visible to upper management and/or tied directly to revenue.

The grind aspect is real though. 99% of FAANG engineers aren't building the next google maps or LLM, they are doing Enterprise CRUD + ELT + jira tickets. Companies like Meta and Amazon have enormous workloads and thus a grinder is preferable.

After a few years of experience, CRUD and ETL can be done while sleepwalking so the only missing component is someone willing to grind, e.g. someone who will spend 100 consecutive days doing leetcode