Should international candidates say "I" or "we" in a US interview?

Both, split by function. Use "I" for decisions you made and work you personally did, and "we" for the scope and the outcome. "We migrated forty services; I owned the rollback path and made the call to dual-write for three weeks." A US interviewer has to defend a hire recommendation to people who never met you, and evidence with no name attached to it cannot be used in that room.

How much self-promotion is too much in an American job interview?

The ceiling is the point where the story stops surviving a follow-up question. Claim the decisions you actually made, attach one number, then name what other people did in a single sentence. Most large US employers score collaboration on its own line, so a candidate who takes credit for a team of six loses there even when the technical answer was strong.

How do you answer "tell me about a time you failed" in a US interview?

Give a real failure, your causal contribution to it, and then a recovery clause: what you changed and the evidence that the change held. Without the recovery clause a genuine self-criticism gets written down as an unresolved risk. A disguised strength such as "I care too much" reads as evasion. Do not name a colleague or a former employer as the cause.

Is it rude to interrupt an interviewer in the United States?

Not in three specific slots: when they are restating your idea incorrectly, when they are explaining a constraint you already have, and when you have realised your own approach is wrong. Say "can I jump in for one second" and be brief. Interrupting the statement of the question itself, or the closing logistics, is where it reads as rude.

What should you say when a US recruiter asks your salary expectations?

Deflect once if you want the band first, then give a range in total compensation, base plus bonus plus equity, and ask what the role is budgeted at. "I am flexible" alone invites a low anchor. A number of US states and cities prohibit asking your salary history, though the list changes, so check your own state rather than assuming.

Does US interview culture vary between companies?

Enough that calibrating to a stereotype is a mistake. A trading desk, a research lab and a fifteen-person startup reward different registers, and many US-based interviewers are immigrants themselves. Read the first five minutes and mirror the interviewer's pace and formality. The one thing that holds almost everywhere is attributable evidence: they need to know which parts were yours.

US Interview Culture for International Candidates: Calibration

An American interview asks you to make a claim about yourself and then evidence it, and the acceptable band is narrower on both sides than it looks. Under-claim and the written feedback says "unclear ownership", which sinks you in a debrief where your interviewer has to defend a stranger to four people who never met you. Over-claim and it lands on the collaboration line instead, which sinks you just as reliably. Below is a calibration table across the five dimensions that catch international candidates out, with the same answer written three ways so you can hear where the middle sits.

Why "we" costs you, and it is not about ego

At most US employers the round does not end when the call does. The interviewer writes a short recommendation, then sits in a debrief where people who did not meet you argue about it. The only currency in that room is evidence attached to a name. If your best story is "the team reduced checkout latency", your advocate has nothing to say when somebody asks what you personally did, and the safe vote in an expensive hiring process is always no.

So the pressure toward "I" is structural rather than a matter of cultural taste. It also has a hard ceiling, because the same scorecard usually carries a separate line for collaboration, and a candidate who claims the work of six people loses there. The band you are aiming for is the register where every claim survives one follow-up question.

The calibration table: the same answer, three ways

Read the middle column out loud. Reading it silently is what makes it feel arrogant. Spoken at normal pace, it does not.

Dimension	Under-calibrated, and how it is recorded	Calibrated for a US panel	Over-calibrated, and how it is recorded
Self-promotion "Tell me about a project you are proud of."	"It was a team effort. We moved billing onto the new ledger and it went well." Recorded as: could not articulate own contribution. Nothing in it is quotable in the debrief.	"We moved billing onto the new ledger over four months. I owned the dual-write and the rollback path, and I made the call to run both ledgers for three weeks instead of one, which caught a rounding bug before any customer saw it. Two other engineers built the reconciliation job."	"I rebuilt billing single-handedly. Before I joined there was no plan." Recorded as: no acknowledgement of the team. It also fails the very next question, because the interviewer will ask who else was on it.
"I" versus "we" Pronoun policy inside that same story	"We" for everything, including the code you typed alone at midnight. Common in candidates trained to treat individual credit as impolite. The interviewer cannot separate you from your old team.	"I" for decisions and actions that were personally yours, "we" for scope and outcome, plus one explicit credit sentence near the end. Roughly: we did the thing, I did these parts of it, they did those parts.	"I" for work six people did. This does not collapse in the story, it collapses in the follow-up, when "what were the others responsible for" gets a vague answer and the whole account becomes suspect.
Failure "Tell me about a time you failed."	"I shipped a migration that took the API down for two hours. It was my fault." Full stop, no recovery. Honest, and recorded as an unresolved risk, because nothing in it shows the same thing will not happen here in month three.	"I took the API down for two hours. I had tested against a copy with a tenth of the rows, so the index build behaved differently under real volume. I wrote the post-mortem, and since then I will not sign off a migration without a load test at production row counts. That caught a similar problem two releases later."	"My real weakness is that I hold myself to impossibly high standards." Recorded as: evasive, no self-awareness. The other over-calibrated form is a real outage blamed on a named colleague or a former employer, which scores against you under accountability.
Salary "What are you looking for?"	"Whatever you think is fair, I am flexible." Reads as no market knowledge, and it invites the recruiter to anchor at the bottom of the band. Deflecting the same question three separate times reads as game-playing.	"I would rather hear the band first, if you have it." If they push: "For this level in this city I am targeting 170 to 190 total compensation, base plus bonus plus equity, and I am flexible on the mix. What is the role budgeted at?" Quote total comp rather than base alone, and set the bottom of your range at a number you would genuinely accept, because that is the number you will be offered.	"I have a competing offer at 250 and I will not go below it," when there is no such offer. US recruiters ask for detail. An invented offer is the one compensation mistake that ends the process rather than merely costing you money.
Interrupting Turn-taking during a live round	Complete listening silence, no verbal acknowledgement, waiting for a full stop that never arrives. In American English the absence of backchannel, the "right", "got it", "mm-hm", reads as confusion, and on a laggy video call it reads as a frozen screen.	Backchannel while they talk. Interrupt in three slots: they are restating your idea incorrectly, they are explaining a constraint you already have, or you have realised your own approach is wrong. "Can I jump in for one second?" then be short. Interrupting yourself is nearly always fine.	Talking over the statement of the question, finishing their sentences for them, or cutting in to correct trivia that does not change the answer. Also cutting into the closing logistics, which is the part where somebody is telling you what happens next.

One structural note sits underneath the interrupting row. American English tends to run on very short turn-taking gaps, and a pause of a second or two gets filled by whoever is quicker. If your first language leaves longer gaps between turns, you will lose the floor repeatedly without anybody intending it. The fix is a two-word placeholder that claims the turn before you have composed the sentence: "Two things." Then think.

Re-cutting one "we" story into an "I" story, in five passes

Write it in your own natural register first

Ninety seconds, "we" and all, no editing. Fighting the pronoun while you are still recalling what happened produces a story with neither accuracy nor ownership. Get the events down, then edit for attribution.

Underline every verb and name the human behind it

Beside each one, write who actually did that thing. Verbs with your name attached become "I". Verbs with somebody else's name stay "we", or better, that person gets named in the credit sentence. This is a mechanical pass, and that is exactly why it holds up under pressure: you are not judging how immodest to be, you are reading off a list you made while calm.

Find the decision, and the option you rejected

American behavioural rubrics are built around judgement rather than activity. "I chose the dual-write over a maintenance window, because a two-hour outage on a Friday would have hit the payout run." A story with no rejected alternative reads as work you were assigned, not a call you made.

Attach one number, then one credit sentence

One number, and it does not need to be impressive, it needs to be real: three weeks, forty services, two hours. Then the credit sentence, which buys back everything the "I" pass spent. Cut it and the identical story starts scoring badly on collaboration.

Say it out loud, timed, until 100 seconds feels short

Ninety to 120 seconds is the shape of a US behavioural answer. Under 40 seconds and the interviewer has to interrogate you for the content, which makes for a bad write-up. Past three minutes and they stop taking notes. Rehearse aloud rather than silently, because articulation is the part that fails when your pulse is up.

Four things the table cannot hold

Small talk is not a test, and skipping it has a cost. The first minute or two of a US call is usually weather, weekend, commute. Going straight to "shall we begin" reads as brusque. One or two sentences back, then let them steer.
"Any questions for us?" is not optional. Having none is frequently written down as low interest. Bring two: one about the work in the next six months, one about the person sitting in front of you. And saying plainly, once, at the end, that you want the job is normal here and does not read as desperate.
Work authorisation questions get a flat, factual answer. If you are asked whether you will need sponsorship now or in future, answer in one sentence and move on. No apology, no essay, no legal analysis of your own case. The details belong with the recruiter, not with the engineer running your technical round.
Salary history and pay ranges are regulated unevenly. A number of US states and cities restrict asking about your salary history, and several require a pay range in the posting. Which jurisdictions, and the exact wording, changes over time, so check your own state rather than assuming. Asking your expectations is a different question from asking your history, and it is permitted everywhere.

Rehearsal is the fix, and where a copilot stops helping

Register is a rehearsal problem, not a live one. Under pressure you produce what you have already said out loud, which is why reading a good answer twenty times does less for you than saying a mediocre one five times. Build the question bank for the specific company and role first, then answer it aloud into a recording. The company Q&A generator exists for that step, and the free tier is five minutes a day with no card, metered only while the copilot is actually receiving interviewer audio rather than for the whole session.

In the round itself, OfferPilot AI transcribes the interviewer on-device with Whisper and drafts a speakable answer grounded in your own resume in roughly a second, which helps when a question lands that you did not prepare for. It reads system playback, so headphones are fine. The Interview Copilot page covers the mechanics.

Where this is the wrong tool

It cannot hear you. The copilot listens to the interviewer, not to you. It has no idea that you said "we" nine times, that your voice rose at the end of every claim, or that you dropped the credit sentence. There is no register meter, and register is the entire subject of this page.
It cannot read the room. Nothing tells you whether this particular interviewer wants the trading-desk register or the research-lab one. That comes from listening to how they talk in the first five minutes and mirroring their pace and formality.
Reading is visible. The overlay is excluded from screen capture at the operating-system level, so it is absent from Zoom, Teams, Meet and screen recordings. That covers screen sharing only. A webcam still shows your eyes tracking text, and in a behavioural round, where the interviewer is watching how you speak, reading is the wrong move whatever the software hides. It also does nothing about a lockdown browser, a proctor, or a person sitting in the room with you.
Windows 10 and 11 only. No macOS, no Linux, no mobile, no browser extension. Plenty of culture-fit conversations and recruiter screens happen on a phone or in person, where no audio routes through the PC and there is nothing for it to hear.
It does not shorten the adjustment. Moving from a register where claiming credit is impolite to one where it is required takes weeks of saying the sentences aloud until they stop feeling like lying. No live tool compresses that, and treating one as a substitute for the rehearsal is how people end up reading somebody else's phrasing in a voice that is audibly not their own.

If you take one thing from the table, take the credit sentence. It is the cheapest line in a US interview: six seconds, it makes the "I" claims around it credible rather than grabby, and it is the sentence most under-calibrated candidates were already trying to say.

US interview culture for international candidates: calibrating how much credit you take