FDE, ML engineer, research engineer, platform — what actually differs day to day?

[akash-coded]

These titles overlap heavily and the job postings are near-identical, which makes it hard to tell
what you are actually interviewing for.

My read, from the shape of the interview loops rather than the postings:

Role	The loop tests	The day is mostly
Forward-deployed / deployment engineer	ambiguity, stakeholders, sequencing — "what is week one?"	figuring out what the client actually needs, then building the smallest thing that proves it
ML engineer	the mechanism, and whether you can measure it	pipelines, evaluation, and the gap between offline and online
Research engineer	falsification — "design the experiment that would disprove this"	making somebody else's result reproduce, and finding out why it does not
Platform / infra	scale, failure modes, cost	the 69% of the bill that is not tokens

The interview sheets go
further on what each round scores.

What I want from this thread

Correct me. If you hold one of these titles, what does the table get wrong? The most useful
reply is the boring detail — what you spend Tuesday morning on.