Sign in to keep reading
“The Room You'll Be In” is free with an account. Your progress and bookmarks follow you across devices.
Sign in to continueNEW: ML Mock & Coaching with a 20+ year experience coach. Save $20 on sessions for Premium.ML Mock & Coaching now available
NEW: ML Mock & Coaching with a 20+ year experience coach. Save $20 on sessions for Premium.ML Mock & Coaching now available
Whose editor, whose model, and what that changes. The two environments this round runs in, why the tooling is usually worse than yours, and what to ask before the day.
“The Room You'll Be In” is free with an account. Your progress and bookmarks follow you across devices.
Sign in to continueThe session you walked through could have run in two quite different rooms, and they fail you in opposite ways. This lesson is the logistics — deliberately last, because none of it matters if the process is not there, and all of it is cheap once it is.
| Two Environments | ||
|---|---|---|
| Their environment | Yours | |
| Editor | A browser platform — CoderPad, HackerRank | Your own, over screen share |
| The code | An existing repository, prepared for the problem | Often greenfield, written from scratch |
| The model | Whatever they provide, sometimes chosen from a list | Whatever you use daily |
| Typically | Meta, LinkedIn, Uber | Shopify, Canva, Rippling, Adobe |
| How it hurts you | Unfamiliar tooling you cannot swap out mid-round | Nowhere to hide — your ordinary workflow is the exhibit |
In their environment, the constraint is that you adapt or you struggle. The model may be older or throttled, autocomplete may behave differently, and it will not have indexed the repository the way your setup indexes yours. Candidates who have only practised in a finely-tuned personal environment discover mid-round how much of their speed belonged to the environment rather than to them.
In yours, the constraint is exposure. Every keystroke, every prompt, every accepted suggestion happens in your natural workflow. If your habit is to accept large generations unread, that habit is now the interview.
The symmetry is worth holding onto: one punishes you for depending on a setup you will not have, the other punishes you for a workflow that was only ever fast. Practising one leaves the other gap open.
If you are handed an empty file rather than a broken one, the session in this course still applies — the phases just compress. There is no bug to reproduce, so Completeness shows up as defining the test before the implementation. There is no existing codebase, so AI Fluency shows up as establishing intent from the spoken brief rather than from a docstring.
What does not change: you are still scored on whether you read what comes back, keep the decisions with trade-offs in them, and can explain what is on your screen.