Pick a project you are proud of. Which decisions were yours alone, and which would have happened without you?

Put your effort into the second half. Anyone can describe a good project; the harder part is saying what would have been different without you. Leo Mehr is Director of Engineering at Ramp, which hires FDEs. He said on a podcast published in July 2026 that he assesses ownership by asking candidates about the decisions they made in their careers and what they were responsible for; he was describing hiring in general, not the FDE loop. The structure below is ours; no employer publishes how it scores this answer. Choose the project by its decisions, not its size. Pick one where you made a call against a reasonable alternative that someone else was pushing. A small project with a clear fork beats a flagship where you were one of many. Give context in three sentences. The goal, who was on the team, and what you owned. Name roles: “a product lead, two backend engineers, and me as the engineer on site with the customer.” Sort the decisions into three bins out loud. Mine: I proposed it, argued for it, and it would have gone the other way without me; say what the other way was. Shared: I shaped it, someone else decided; name them by role. Inevitable: the deadline, the stack, the customer’s hard requirement. The third bin is what makes the first believable. For each of your decisions, give the alternative, your reason, and the evidence that it was right, or that it was not. Close with the counterfactual in one sentence. “Without me it would still have shipped; it would have shipped with X, and Y would not exist.” Say “that was my call” and “that was the product lead’s call; I brought the data.” The trap runs both ways: claiming everything collapses under the follow-ups, and “it was a team effort” gives the interviewer nothing to evaluate. Auditing which decisions in your own projects were yours, and turning them into stories, is taught in Behavioral mastery for FDEs, in Pro. To practice the long version of this answer, try the project deep dive.

Which decisions were yours alone? | FDE interview question

A practice prompt we wrote. No company or candidate report names it, so it carries no company tag.

How to answer

Put your effort into the second half. Anyone can describe a good project; the harder part is saying what would have been different without you. Leo Mehr is Director of Engineering at Ramp, which hires FDEs. Source 1Forward Deployed Engineering (Leo Mehr, Director, Engineering)PublisherRamp Builders (engineering blog)Source typecompany blog He said on a podcast published in July 2026 that he assesses ownership by asking candidates about the decisions they made in their careers and what they were responsible for; he was describing hiring in general, not the loop. Source 2Leo Mehr - Ramp's $44B Bet on Services (YouTube auto-generated English captions)PublisherBasil Chatha, YouTubeSource typerecorded talk or interview The structure below is ours; no employer publishes how it scores this answer.

Choose the project by its decisions, not its size. Pick one where you made a call against a reasonable alternative that someone else was pushing. A small project with a clear fork beats a flagship where you were one of many.
Give context in three sentences. The goal, who was on the team, and what you owned. Name roles: “a product lead, two backend engineers, and me as the engineer on site with the customer.”
Sort the decisions into three bins out loud. Mine: I proposed it, argued for it, and it would have gone the other way without me; say what the other way was. Shared: I shaped it, someone else decided; name them by role. Inevitable: the deadline, the stack, the customer’s hard requirement. The third bin is what makes the first believable.
For each of your decisions, give the alternative, your reason, and the evidence that it was right, or that it was not.
Close with the counterfactual in one sentence. “Without me it would still have shipped; it would have shipped with X, and Y would not exist.”

Say “that was my call” and “that was the product lead’s call; I brought the data.” The trap runs both ways: claiming everything collapses under the follow-ups, and “it was a team effort” gives the interviewer nothing to evaluate.

Auditing which decisions in your own projects were yours, and turning them into stories, is taught in Behavioral mastery for FDEs, in Pro. To practice the long version of this answer, try the project deep dive.

Follow-ups

What the interviewer may ask next, once your first answer is on the table.

The person who disagreed with you on that call. What did they think, and were they partly right?
Which of the decisions you called yours turned out to be wrong?
If I asked your manager from that project what you owned, what would they say?

Where answers go wrong

Saying “we” for every decision, so the interviewer cannot find anything to credit to you.
Claiming calls that belonged to a manager or the customer. The claim falls apart on the first follow-up.
Spending the opening on the customer’s business, so the decisions arrive when time is nearly up.

Answer this in two minutes

Write the answer you would say out loud. The clock starts with your first word.

Compare with the model answer

Illustrative answer about a fictional project

Written in the first person to show the structure. Tell yours from your own work.

The project I’d pick is a dispatch tool I built for a regional freight carrier while I was the deployed engineer at a logistics software company. Their dispatchers matched drivers to loads from a shared spreadsheet, and it took about 11 minutes per load at peak. The team was a product lead, two backend engineers in our office, and me on site with the carrier three days a week.

Three decisions were mine.

First, what we shipped first. The product lead wanted automatic assignment in the first release, because that was what the carrier had bought. I argued for a read-only view that showed our top three drivers for each load next to the dispatcher’s own choice, and for running it in shadow for two weeks. My reason was that the dispatchers had already rejected one vendor tool, and I had watched them override it. Scope was hers, so arguing wasn’t enough. I offered a two-week shadow run with a fixed date to decide, and she took it because it cost her nothing if I was wrong. Without me, auto-assignment ships first. The shadow run showed our pick matched theirs 64% of the time, and the mismatches mattered more than the matches.

Second, what the mismatches meant. I read all 212 override notes myself instead of treating them as noise. Most of them came down to drivers’ remaining legal driving hours, which our model did not see. I proposed making hours a hard filter rather than a scoring feature, and I wrote that filter against the carrier’s electronic logging feed. That constraint would not exist without me, because nobody else was reading the notes. Once hours were a hard filter, overrides fell from about one load in three to one in eight.

Third, a call that went wrong. I keyed loads on the carrier’s depot codes instead of building a mapping table, because the depots I had checked all used the same codes and a table would have cost a week. I didn’t check the other three. One of them used its own codes, and for five days its loads never showed up in the tool, so its dispatchers went back to the spreadsheet and nobody told us. I found it from a gap in the daily counts. I told the head of dispatch before he asked, built the mapping table, and made the nightly import fail on an unknown code instead of skipping it.

Two decisions were shared. The data model was the senior backend engineer’s; I pushed for keeping the raw logging events rather than a daily summary, and she agreed after I showed her two override cases that needed them. The rollout order across depots was the head of dispatch’s call; I gave him the shadow numbers per depot.

Some things would have happened anyway: the move onto the carrier’s cloud account, Postgres, and the deadline before their peak season.

We went live with assisted assignment in eight weeks. Time per load, measured from the load-created and driver-assigned timestamps, fell from 11 minutes to about 4. Without me it would still have shipped, as auto-assignment with no hours filter, and my read is that the dispatchers would have overridden it the way they overrode the last tool.

Interviewer: If I asked your manager from that project what you owned, what would she say?

Candidate: The shadow run and the hours filter, and the depot-code miss, because I told her the day I found it. She’d say the data model was the senior backend engineer’s, and that I pushed to keep the raw logging events in it.