Staff Engineer Interviews: What Changes at Staff+

The staff engineer interview has mostly the same rounds as the senior loop you have already passed, which is what makes it easy to misjudge. The calendar invite still says coding, system design, behavioural. What changed is the sentence each interviewer is trying to write afterwards. At senior level it is "can own a hard project end to end". At staff it is closer to "changes what several teams build, and they are glad of it".
Candidates who miss that sentence often get an offer one level down instead of a rejection, with feedback such as "strong senior, not enough scope". Nearly always the cause is the evidence they chose to present: excellent execution of work that someone else framed. This guide goes round by round through what staff-level evidence looks like, with most of the space given to the project deep dive, because that round has no senior equivalent and cannot be prepared the way the others can.
What staff-level scope means to the people grading you
Will Larson's StaffEng describes four staff archetypes (tech lead, architect, solver, right hand), and companies use their own names, but the interviewers are listening for the same four things whichever one you are.
- Scope. The work changed something for several teams or a whole organisation, and you can say which teams and what changed for each.
- Ambiguity. You were handed a problem before it was a project, often a problem that nobody had framed yet, and the framing was your contribution.
- Influence without authority. Other teams changed their plans because of a document you wrote, a prototype you built or an argument you won, with no reporting line to lean on.
- Judgment at organisational stakes. You made calls that would have cost engineer-years if wrong: a platform bet, a migration strategy, a build or buy decision. You can explain the alternatives you turned down and what you would do differently now.
Scope is not the same as size. A story about a system with enormous traffic can still be a senior story if your part was one well-run component. A story about a dull internal migration can be a staff story if you were the person who decided it should happen, got five teams to agree to the order, and stopped it when the plan turned out to be wrong for two of them.
The loop from recruiter call to levelling decision
Formats differ by company, and none of what follows is a published standard, so ask your recruiter for the list of rounds and which of them feed the level decision. A common shape is a recruiter call, one or two screens (a coding or design round, or a conversation with the hiring manager), then a loop of four to six interviews. What differs from the senior loop is where the weight sits.
| Round | What a senior answer shows | What a staff answer adds |
|---|---|---|
| Coding | Clean, tested solution while talking | Little extra beyond independence; a miss can still sink the loop |
| System design | A sound architecture for the stated problem | Reshapes the problem, names organisational constraints, goes deep in two areas |
| Project deep dive | Often not asked at senior level | Sixty minutes on your own work: decisions, alternatives, cost, what broke |
| Behavioural or leadership | Ownership, conflict handled well | Direction set across teams, disagreement with senior people, mistakes at scale |
| Hiring manager or cross-level | Fit for the team | Whether a director would hand you an unframed problem |
Two practical consequences follow. First, the coding round cannot earn you the level, so preparing for a staff loop the way you prepared for a senior one (mostly problems) puts the hours in the round with the least to give. Second, the level is usually decided after the loop, by a debrief or a committee reading written feedback, from the scope of the evidence across all rounds. No single interviewer awards it, which means every round needs at least one moment that reads as staff.
The project deep dive
Different companies call it the project retrospective, the technical deep dive or the experience round. The format is one interviewer, sometimes two, and forty-five to sixty minutes on a single piece of work you led. It sounds like the easy round because you know the material. It is the round where interviewers find the edge of your actual contribution, and they have a full hour to look for it.
Choosing the project
Pick it weeks ahead. Three tests help.
- Your fingerprint is on the decisions, not only on the code. You can point to the two or three calls that shaped the outcome and say they were yours.
- It crossed a boundary. Other teams, another function, a vendor, a compliance constraint. A project that lived inside your own team gives the interviewer nothing to write about influence.
- Something went wrong. A project with no surprises reads either as small or as edited.
The best candidate is rarely your most recent project. Recent work is often unfinished, and without an outcome there is no result to defend. It is also rarely the one with the biggest numbers if your role in it was narrow.
How the hour usually runs
This is an example outline for a sixty-minute deep dive, written for this guide. It is not any company's script.
- Minutes 0 to 5. You set the scene: the business problem, the state of the system, the teams involved, and your role in one sentence. Draw the system if there is a whiteboard or a shared canvas.
- Minutes 5 to 20. The interviewer picks the first decision and walks down it. Why this approach? What else was on the table? Who disagreed, and what was their argument?
- Minutes 20 to 40. Depth. They choose one component and keep asking until you reach something you do not know. Data model, failure handling, the migration order, how you measured.
- Minutes 40 to 55. Outcome and hindsight. What did it cost, what broke after launch, what you would redo, what happened to the system after you left it.
- Last 5 minutes. Your questions.
The follow-ups to rehearse
These are example questions of the kind interviewers ask in this round:
- What was the strongest argument against the direction you chose, and who made it?
- Which part of this did you personally design, and which parts did you review?
- What would have happened if you had done nothing for six months?
- Where did the estimate go wrong, and when did you know?
- What did this cost in engineer time, and was it worth that?
- Which decision would you reverse today?
- How did the teams who had to adopt it find out, and what did they have to stop doing?
The honest boundary is the strongest answer to the second question. "I designed the consistency model and the rollout order; the storage layer was another engineer's design, which I reviewed and pushed back on in two places" tells the interviewer exactly where your judgment was exercised. A candidate who claims the whole system gets asked about the part they know least.
Preparing it
Write a two-page document on the project as though it were a design review written afterwards: context, options considered, the decision, the outcome with numbers, what you would change. Then have someone ask you questions about it for forty-five minutes. You cannot rehearse this round alone, because the part that matters is the fourth follow-up on a topic you did not pick. A peer at staff level who was not on the project is ideal. If you have nobody suitable, a session with one of the coaches who work with Staff+ candidates covers the same ground, with the advantage that the person asking has sat on the other side of this round.
System design at staff altitude
The prompt is the same one a senior candidate gets. The grading moves from coverage to judgment, and it shows in three places.
The problem gets reshaped before the architecture starts. A senior candidate asks about scale and features, then designs. A staff candidate also asks what already exists, who operates it, what the migration path from the current system is, and which requirement is negotiable. "Does this need to be real time, or does the business need it to feel fresh within a minute?" changes the design more than any technology choice later.
Organisational constraints are named. Real systems are built by teams. Saying "I would split this along the boundary between the ingestion team and the serving team, so the contract between them is this queue and this schema" shows you have lived with the consequences of a design, and that you know architecture and team structure shape each other.
Depth is uneven on purpose. You will not cover everything in forty-five minutes. Say so, pick the one or two components where the risk sits, and go further than the interviewer expects: the failure modes, the numbers, what you would measure in the first month. Then say what you are deliberately leaving shallow and why.
An example of the difference, invented for illustration. Asked to design a notification service, a strong senior answer produces queues, workers, a preference store and retry logic. A staff answer does that in twenty minutes and spends the rest on the hard part it identified early: every product team will want to send, so the system needs a policy layer for rate limits and priorities, an owner for that policy, and a plan for moving forty existing senders onto it without a flag day.
The framework for system design interviews still applies, and the guide to system design interview questions by level shows the same prompt answered at senior and at staff.
Coding at staff level
Most staff loops keep at least one coding round, and it is graded at roughly the senior bar. Nobody expects harder algorithms from you. They do expect that you have not let the skill lapse.
It catches staff candidates for a plain reason: several years of design documents and reviews, and little timed coding. Two weeks of problems done aloud, against a timer, in the language you will use, is normally enough to get the fluency back. Ask the recruiter whether the round allows AI tools, because formats have been changing; the guide to AI in coding interviews explains what to ask.
Behavioural signals in a staff loop
The behavioural round at staff level is an influence round. The questions sound familiar (a disagreement, a failure, a project you are proud of), but the interviewer stays on one story for fifteen or twenty minutes and listens for the mechanics: what you wrote, who you had to persuade, what they said, what you gave up to get agreement.
Four kinds of story cover most of what gets asked:
- A technical direction you set that other teams adopted.
- A project you stopped, possibly your own.
- A disagreement with someone more senior than you, and how it ended.
- A call you got wrong at organisational scale, who paid for it, and what you changed.
The last one matters more than candidates expect. At this level, a career with no expensive mistakes reads as a career with no large decisions in it. Give the cost plainly and spend the time on what you changed afterwards.
The bank of staff engineer behavioral interview questions has 35 questions grouped by signal, with one story told at senior and at staff level side by side. The STAR method still gives the shape, with most of the time spent on the actions.
The usual reasons for a down-level
A down-level means the debrief agreed you should be hired and did not agree on the level. These are the patterns behind it.
- Senior-shaped stories. Every example is a project delivered well inside one team. The fix is selection: you probably have the cross-team stories and did not choose them.
- "We" throughout. The interviewer cannot separate your decisions from the team's. Say "I" for decisions and "we" for delivery.
- Depth that runs out. In the deep dive, the third follow-up reaches a layer you only reviewed. Choose a project where your knowledge goes to the bottom of at least one component.
- Design at the wrong altitude. A correct architecture, with no reshaping of the problem and no opinion about what is risky.
- No visible disagreement. Staff engineers are expected to have pushed back on someone senior at least once. A candidate with no such story looks like someone who implements other people's direction.
- One weak round. A coding miss or a thin behavioural round removes the margin, and doubt about level tends to be resolved downwards.
- A mismatch in role shape. Your experience is deep in one system and the role needs breadth across many, or the reverse. This one is about fit and no preparation changes it.
If it has already happened, the guide to a down-levelled offer covers how the decision is made, the four ways to respond and the money over four years.
The level conversation with your recruiter
Have it before the loop. Ask which level the loop is calibrated for, whether the company will consider the level below if the evidence falls short, and which rounds carry the level decision. Recruiters generally answer these questions, and the answers tell you where to put your preparation.
After the loop, the level is the first item of the negotiation, ahead of any number. A level sets the band, the equity and the scope you will be given in the first year, and a counter-offer inside the lower band rarely makes up the difference. The salary negotiation guide covers the mechanics.
Where coaching fits
The parts of a staff loop that decide the level are the parts you cannot check alone: whether your deep dive survives an hour of follow-ups, and whether your stories read as staff to someone who has made that call in a debrief. On techinterview.coach, 20 of the 40 coaches list leadership coaching as of November 2026, among them Staff and Principal engineers and engineering managers. The leadership coaching page shows their backgrounds, so you can choose someone who has interviewed at the level you are targeting and ask for a mock deep dive on the project you plan to present. For repetition between sessions, the practice interview runs a behavioural or system design round by voice with an AI interviewer and a written recap; it is free, sign in to start.
Related guides
- Staff engineer behavioral interview questions: 35 questions by signal, for the round this guide only summarises.
- Down-levelled offer: what to do when the loop ends one level below the one you interviewed for.
- System design interview questions by level: the same prompts answered at mid, senior and staff.
- A 4-week prep plan for senior engineers: the weekly schedule, with the changes to make for a staff loop.
- Principal engineer interview: the level above, and what its loop adds to this one.
- The engineering manager interview: the other route at this level, if you are weighing both.