1 min lesson
Why the flat org shapes the loop
Talk this through in your own words: "The JD says Cursor has 'no narrow job descriptions.' What is the most accurate read of that line for this role?" Finish with the next move.
Step 1 of 2
Model behavior, retrieval, agent loops.
You're expected to have opinions about what models can and can't do yet.
UX for workflows that didn't exist a year ago.
Taste and polish are part of the engineering, not a separate review gate.
Shipping fast, correct, low-latency code.
End-to-end across the stack, no handoff that finishes your work.
The interview is engineered to test full ownership because the job is full ownership. The eight-hour build-on-our-codebase onsite hands you a vague prompt and watches whether you can go from nothing to a shipped feature without a manager breaking the work into tickets. If you read the org as flat, you should expect to be tested on autonomy and you'd be right.
When the recruiter or hiring manager asks what you think this role is, lead with ownership of an outcome on the editor, then name the multi-hat reality. Say you expect to write the agent code, watch real users hit it and tune the copy that explains it. That framing signals you understood the flat-team JD, while candidates who describe a narrow backend role signal they didn't.