Forward Deployed Engineer interview questionsBehavioral — one engagement, end to end round

Purpose

Sets the three constraints that make the hour comparable across candidates - embedded, running against the customer's systems, finished - and announces the closing question at the start rather than springing it at the end. Saying it early is the point: a candidate who spends the hour knowing they will be asked for the view of someone who was there has every reason to keep the facts straight, and it costs nothing to say. The question asks for a role and a criticism, never a name or a contact: a candidate's former customers are not yours to call.

Purpose

A calibration question with a real read in it. Someone who ran the engagement can compress it; someone who watched it tends to narrate it in order and never reach the outcome. It also fixes the facts you will check the rest of the hour against, so write them down.

Signals to score

• Covers customer, problem, what shipped, when and their own part, inside about two minutes
• States the customer's problem in the customer's units - money, hours, risk, a date - rather than as a technology
• Gives the engagement's length and dates without being asked
• Names their own part precisely, including who else worked on it and what those people did
• Describes what shipped as something that ran, not as a recommendation, a prototype or a document
• Names the outcome the customer judged it on, even if it was missed
• Does not open with the architecture

Follow-up questions

• What would the customer say the problem was, in their own words?
• How long were you on it, start to finish?
• Who else worked on it, and what did they do?
• What number did the customer care about?
• Did it hit that number?

Purpose

The core discovery read. Almost every real engagement has a gap between the ask and the problem; the question is whether the candidate found it, how early, from what evidence, and whether the customer agreed to the change in a way that would survive a disagreement later. A candidate who says there was no gap either had the rare textbook engagement or built what was asked for, which is the failure the role exists to prevent.

Signals to score

• Gives the original ask in the customer's words, and keeps it separate from their own later reframing
• Names the gap between the ask and what was built in a sentence or two
• Says when the gap was found, in days or weeks from the start
• Says what evidence showed it - something watched, counted or read - rather than an impression
• Names who at the customer agreed to the change, by role, and how the agreement was recorded
• Separates the person who paid for the work from the person whose work changed, and says what each wanted
• Says what would have happened if they had built the original ask
• Admits a part of the ask they misread, or found too late
• Keeps "I" and "we" distinct, and can say which parts of the discovery were theirs

Follow-up questions

• If you had built exactly what they asked for, what would have happened?
• What did you see that made you believe the ask was wrong?
• How many days in was that?
• Who agreed to the change, and where is that written down?
• Who at the customer disagreed?

Purpose

The first week decides most engagements, and it is the hardest part of the job to fake in detail, because it is made of small specific decisions - who to meet before whom, what to ask for on the first day. Candidates who have only read about the role describe a discovery framework. Candidates who have done it describe a calendar. A candidate who has prepared for this question, or met a discovery exercise earlier in the process, will know which answers sound right; the calendar - days, people, what arrived when - is the part preparation cannot supply, so score that.

Signals to score

• Names the people met in the first week by role, in an order with a reason behind it
• Met the people who do the work, not only the sponsor and the IT contact
• Watched the work happen, read real records or ran a query on real data inside the first week
• Asked for access in the first days, and says how long it actually took to arrive
• Asked what had been tried before, and says what that changed
• Names a number established in the first week that they still used at the end
• Sent the customer something written by the end of the week that the customer could correct
• Names something they deliberately left until later, and why

Follow-up questions

• Who did you not get to meet, and did it matter?
• What did you ask for access to on the first day, and when did it arrive?
• What did the users tell you that the sponsor had not?
• What did you send the customer at the end of that week?
• If you ran that week again, who would you meet first?

Purpose

Separates a trade from a slip. Every engagement drops something; the read is whether it was dropped by decision - priced, put to the customer, agreed - or by running out of time and letting the demo break the news.

Signals to score

• Names a specific thing that was dropped or deferred, and who had asked for it
• Priced it at the time, in days, people or money, rather than in hindsight
• Put the choice to the customer explicitly, and the customer decided
• The person who wanted it heard directly, before the launch rather than at it
• Offered a smaller version or a later date, and says whether it was taken
• Says plainly what leaving it out cost the customer
• Knows whether they said "no" or "not yet", and chose deliberately
• Did not let the schedule make the cut silently

Follow-up questions

• Did they agree, or did they stop arguing?
• Who told the person who wanted it?
• What did you offer instead?
• Did it ever get built?
• Looking back, what would you have cut instead?

Purpose

The role puts an engineer in rooms where bad news has to be said by the person who knows it, without an account manager to soften it. The read is whether the candidate has done that - early, to the right person, with evidence and an option - and whether the relationship survived. Accept any material piece of news: a date, a number, a risk, a person who will not cooperate, a system that cannot do what was sold.

Signals to score

• The news was material - a date, a number, a risk, a person - rather than a preference
• Names the person who heard it by role, and why it had to be that person
• Said it early, before the customer found out another way, and can say how long they had known
• Gives the words they used, or close to them
• Brought evidence and at least one option along with the news
• Did not route it through their own account or sales team to avoid saying it, or can say why that channel was right
• Says what it cost in the relationship and how that was repaired
• Something concrete changed afterwards: a plan, a date, a scope, a person
• Tells it without making the customer the villain

Follow-up questions

• How long had you known before you said it?
• Why you, and not your account lead?
• What did they say back?
• What would you say differently now?
• Did they trust you more afterwards, or less - and how do you know?

Purpose

Most candidates can say what they handed over. Fewer can say what survived, and fewer still know because they checked rather than because nobody complained. The "how do you know" half is the part being scored.

Signals to score

• Names what was still running, specifically, and what had been switched off or abandoned
• Names the role at the customer who owned it after the handback
• Knows because they checked - a message, a metric, a visit - and says when
• Describes the handback as something the customer did while they watched, before they left
• Names something the customer's side kept and could change: a repository, a runbook, a check that fails into a queue someone reads
• Gives a usage or outcome figure from after they left, or says plainly that they have none
• Is candid about what did not survive, and why
• Distinguishes handing over from handing off

Follow-up questions

• When did you last hear from them, and about what?
• If I called the person who owns it now, what would they say?
• After you went, who could deploy a change?
• What did you leave behind that you would not leave again?
• What would have had to be true for it to still be running a year later?

Purpose

Something always breaks. A candidate who says nothing did was either not told or not looking, and either way did not own it. The read is whether the failure reached the customer's people through something the candidate built, and whether the candidate separates their bug from their handback.

Signals to score

• Has an answer - something broke - rather than claiming nothing did
• Says how the failure was detected, and whether by a check they built or a user who noticed
• The alert, if there was one, reached someone at the customer and not only the candidate
• Owns the part of the cause that was theirs before describing anyone else's
• Says whether it was a defect in the code or a gap in the handback
• Says what they changed so the same kind of failure would not need them again
• Is honest about whether they were still the one being called after they rotated off
• Names something they now do in every handback because of it

Follow-up questions

• Who got the call?
• How long was it broken before anyone noticed?
• Was that your bug, or your handback?
• What did you change on the next engagement because of it?
• Are they still calling you?

Purpose

Announced in the opening line, which is what gives it force: a candidate who has spent the hour knowing it was coming has had every reason to be accurate. Ask for a role, never a name or a contact. The read is the second half: someone who owned an engagement knows what the customer would criticise, because they were told.

Signals to score

• Names a role at the customer who saw the work closely, not only the sponsor who signed it off
• Says what that person would criticise, specifically, without being pushed
• The criticism is about something the candidate did, not about the customer or the vendor
• Says what they changed because of it
• Stays consistent with the rest of the hour, with no new version of an earlier story
• Treats the question as fair rather than as a threat
• Can say briefly what the same person would praise, and it is not simply what they led with an hour ago

Follow-up questions

• Why that person and not the sponsor?
• What would they say first?
• Did they ever tell you directly?
• What did you do about it?
• What would the sponsor say instead?