Question 4Integrations and APIs
The customer is on the call. Their integration is failing on live data, you've never seen their system, and they want it working before the call ends. What do you do?
FDEs debug in front of the customer. This tests whether you stay methodical with someone watching.
“Help lenders adopt new APIs and troubleshoot their API and LOS integrations, building tooling and templates so each new integration takes less effort than the last.”Solutions Engineer · Friday Harbor
The answerFour steps
- 1
Reproduce it on their data first, with them watching
Say it like this
“I ask them to share their screen and pick one order that's missing. We resend it together while I watch our logs, so we both see the same failure.”
Why it scores
Their data, their screen. Both sides see the same failure, so nobody argues about whose bug it is.
- 2
Split your side from theirs: what you sent, what they received
Say it like this
“Our log shows their endpoint answered 200, so the request left us and they received it. That moves the search to their side: their handler log shows the record dropped because a new discount field is null on these orders.”
Why it scores
Uses the hand-off point, what you sent and what they got, to cut the problem in half before reading anyone's code.
- 3
Ship the smallest fix they can see working
Say it like this
“The smallest fix is on our side: send an empty string instead of null for that field, behind a flag for this one customer. I deploy it, we resend the missing orders, and they watch them appear.”
Why it scores
The smallest change the customer can watch working, limited to one customer so nothing else moves.
- 4
Say what you won't touch on this call, and when you'll follow up
Say it like this
“Before we hang up I say what I'm not doing on this call: changing their validation code or backfilling last week's orders. I send a short write-up today and we agree on the proper fix with their team.”
Why it scores
Says what stays out of scope. On a live call, that keeps a quick fix from turning into a risky one.
StrongAll together
About a minute, out loud. The marked parts are the four steps; hover one to find it above.
Say it's our webhook into their order system, and since this morning some orders never arrive. I ask them to share their screen and pick one order that's missing. We resend it together while I watch our logs, so we both see the same failure.1 Our log shows their endpoint answered 200, so the request left us and they received it. That moves the search to their side: their handler log shows the record dropped because a new discount field is null on these orders.2 The smallest fix is on our side: send an empty string instead of null for that field, behind a flag for this one customer. I deploy it, we resend the missing orders, and they watch them appear.3 Before we hang up I say what I'm not doing on this call: changing their validation code or backfilling last week's orders. I send a short write-up today and we agree on the proper fix with their team.4
WeakThe answer that fails
Sounds fine. Common. Scores low.
“I'd ask for their API docs and log access, set up a copy of their system locally so I can reproduce it properly, then write a fix and book a follow-up call to deploy it.”
Interviewer's note
Sensible in a normal week, wrong on this call. The customer watches you leave with nothing working and no idea what broke.
PracticeTwo minutes, then they push back
Your turn
The customer is on the call. Their integration is failing on live data, you've never seen their system, and they want it working before the call ends. What do you do?
Answer like you're in the room: about two minutes, no notes. Say it out loud, even alone. It feels different from thinking it. Then check your answer against what interviewers score.