Interview prep
We send this page to everyone we interview. None of it is a secret we are trading away: the pipeline runs the same eight stages for every candidate on every role, and knowing them in advance is what makes the session about your engineering instead of about guessing what we want to hear.
Stage by stage
Four of the eight stages are ours to run and you never see them. These are the ones you are in, and what the others are doing while you wait.
01You
You entered against one specific role, not a general pool. Everything after this measures you against that role’s requirement list, which is why a strong engineer can be a no here and a yes on the next opening.
02Our system
Your CV is read against those requirements. The screen never rejects anyone on its own; it marks what the interview has to probe and flags the claims worth testing. Thin detail here becomes a harder question later.
03NKE
A principal architect reads everyone the screen marked worth seeing and decides who moves forward. A person makes this call every time; our system is not allowed to make it.
04Our system
You get a link and pick your own slot. Choose a time you can be somewhere quiet, with a stable connection, for the whole session. Rescheduling is cheap; salvaging a bad connection is not.
05NKE
A generated script drives the session: timeboxed sections, your stack probed one technology at a time, and questions drawn from your own history. Notes are written as you talk. That is the format, not a reaction to your answers.
06Our system
Your notes and CV become a starred scorecard, a requirement by requirement fit assessment, your standout projects, a standardized resume, and a profile with your name on it. You are represented by the assessment, not by how your CV was formatted.
07NKE
Nothing reaches a client automatically. Your profile goes out when we would put our own name on it. That gate is the reason the client takes the next meeting seriously when it opens.
08The client
You meet the client’s team yourself, and they arrive holding your scorecard and profile. They already know where you are strong, so the conversation starts further along than a first interview usually does.
Preparing
Not interview theatre. These are what separate the sessions that go well from the ones that stall, in the format we actually run.
Every session pulls questions from your history. The three roles in fifteen months, the migration you led, the number you put next to a result: expect to unpack any of it. If you cannot reconstruct a claim, saying so costs you far less than improvising around it.
Choose something you architected or debugged, not something impressive you happened to be near. The questions are why, not what: the option you rejected, the thing that broke in production, what you would build differently now.
The stack is probed one piece at a time. React, Node, Java, Python, Postgres, Elasticsearch, cloud: whatever the role runs on gets its own questions. "Full stack" is not an answer to any of them.
There is an AI section in every session, and it is about practice rather than vocabulary. What you use daily, where it fails you, what you stopped using and why, how you verify what it produces. Tools you tried once read as tools you tried once.
Part of the session moves to English for everyone. Nobody is scored on accent; the question is whether you can hold a technical conversation with a US team when the subject gets hard. Say two of your project stories out loud in English beforehand.
Notice period, availability, vacation already booked, where you work from, and where you think you sit against the rate in the posting. These are captured as part of your profile, and stating them in the session is much better for you than renegotiating them once an offer is in motion.
How to answer
The most common thing we correct in a session is not a knowledge gap. It is scope. "Have you used React?" is a yes or no question, and the answer to it is one word plus a line of context. If the project behind that yes matters, the next question will ask for the project.
Sections are timeboxed, so three minutes spent on a question that needed ten seconds comes straight out of the questions where your depth would have counted. Clear, precise, concise, in that order.
Some questions are asking for one outright: how you handled a production incident, a disagreement with a lead, a migration that went wrong. STAR keeps those answers to about two minutes and stops the part we are actually assessing from being the part that gets skipped.
What is measured
A process that asks everyone identical questions cannot tell a client whether you fit their stack. One that improvises cannot compare you to anybody. Ours holds the structure constant and generates the content from the role and from your history, so what you get rated on is the work you have actually done.
The session
Most of the anxiety in a technical interview comes from not knowing the shape of it. Here is the shape.
Said plainly
Four patterns sit behind most of the assessments we end up not releasing. They are here once, in one place, because knowing them beforehand is more useful to you than encouragement is.
A result on the CV that cannot survive two follow-up questions costs more than never listing it. Nobody expects perfect recall of a project from four years ago; the question is whether you were close enough to the work to explain how it went.
Naming models and frameworks is not the same as having shipped with them, and the AI section finds the difference in about two questions. The gap does more damage than an honest "I have not built that yet, here is the closest thing I have."
Fluency with a library and no account of what sits underneath it shows up the moment a probe goes one level down. Senior roles are priced on that level, so it is the level the questions aim at.
Conversational English that breaks down once the subject gets specific is the most common reason a genuinely strong engineer is not released. It is also the most fixable thing on this list, if you practise before the session rather than during it.
Questions
A principal architect at NKE. Someone who writes production code that week and would have to work alongside whoever gets placed. The technical bar means something because the person setting it does the work.
Every session includes a mandatory section in English, for every candidate on every role, because the placement means working directly with a US team. The rest of the session follows whatever language you and the interviewer open in.
No deck, no take-home, nothing to submit beyond the CV you already sent. The eight stages are the entire process. The preparation that helps is being able to talk about your own work in detail.
Parts of the pipeline are automated: the CV screen, the scheduling, and the synthesis of interview notes into a profile. None of them accept or reject anyone. Every decision that moves you forward or stops you is made by an architect.
You were measured against one role’s requirement list, and there is no general pool behind it. A no there is a no for that role, not a rating of you as an engineer.
Engineer rates run from $25 to $31 an hour, and the role you applied to published its rate in the posting. Where you land inside that band depends on your profile and on how the process goes, so the session is part of how the number is set. The $37 to $45 quoted elsewhere on this site is what a client pays NKE, which carries the US contract, the insurance and the IP assignment behind every placement.
Before your session
Ask before the session rather than spending its first ten minutes on logistics. We would rather use that time on your engineering.