Blog

How to build a leadership story as a software engineer

"Tell me about a time you led without authority." If that question stops you because you have never managed anyone, this post is for you. In about ten minutes of talking, you can have a leadership story for your next behavioral interview, built from your own work.

By Koaches9 min read

You may think you have no leadership story because nobody has ever reported to you. So you open an empty document, and what comes out sounds like a performance review.

Do it the other way around: talk first and write last. Below are the questions to answer out loud, one story worked from first answer to final version, and the follow-ups it has to survive.

Video

The session behind this post

We recorded this session answering the questions below as a software engineer, then wrote this post from it. Watch it first, or come back to it after the worked story.

You already have one

Leadership in an interview is not about your title. Google's re:Work guide to structured interviewing sets different expectations by level. For people managers, it looks for someone who can empower a team and shape a long-term vision. For individual contributors, it looks for "emergent leadership":

"the ability to take initiative, work collaboratively across teams, and a consistent drive for self-development."Google re:Work, guide to structured interviewing

That is ordinary engineering work. What you are looking for is a time you saw something others did not, you could not order anyone to act, and people did something differently because of you. A few examples to jog your memory:

  • A design nobody asked for. You wrote a proposal to change how two services talk to each other, and got the other team, who do not report to your manager, to agree to the new contract.
  • The alert that kept waking people up. You pulled the last few weeks of pages, showed the team that most had the same cause, and got the fix ranked ahead of feature work.
  • The migration nobody finished. It sat almost done because nobody owned the last few callers. You went to those teams, paired with them and got it over the line.
  • The disagreement you settled with evidence. A senior engineer rejected your rollout plan. Instead of arguing in the comments, you ran a small test behind a flag and brought back the results.
  • The incident nobody was running. Everyone was debugging and nobody was coordinating. You started assigning work, wrote the review afterwards, and got the deploy process changed.

Pick the one from your own work where somebody needed convincing. It does not have to be big, but it has to have happened: one time, with particular people. "I led the payments project" is a job description. "I got the payments team to drop their homegrown retry library" is a story.

Talk first, write last

A story typed from nothing comes out polished and empty. A story you talk through comes out messy and full of detail, and the detail is what makes it yours.

Give yourself ten minutes. Record yourself on your phone, or ask a friend to read you these questions, and answer them in order. Do not fix your sentences. Say what comes to mind.

  1. Think of a time you got people behind a direction you set, with or without authority. What was going on?
  2. What did you do to get the skeptics to take it seriously?
  3. What changed after that in the way people worked day to day?
  4. What changed in concrete terms?
  5. What did you believe before this that turned out to be wrong?
  6. What did you do to make the change last instead of fading out?
  7. What pushback did you get from people who did not want to change, and how did you handle it?
  8. What was the hardest decision you had to make to keep it moving?

The first five give you the outline. The last three give you the resistance and what you did to keep the change going, which belong in the story too, plus the detail you will need when the interviewer starts pushing. If you do not have a number to hand, say so and keep talking. You can look it up afterwards.

One story, from first answer to final version

Here is a session we recorded, answering these questions as a software engineer on a team. The engineer and the results are an example. They show how the method works, and they are not a claim about what AI tools do for a team.

This is what some of the raw answers sounded like:

Figure 1

What was said

What did you do to get the skeptics to take it seriously?"I actually built a feature that I was responsible for and then got it reviewed and then got it merged. And when it was done, I told the team that I've used AI to write every single line of code. And that took the team by surprise."
What changed in concrete terms?"We saw about 20% improvement in terms of PR output that we were generating. And so writing code stopped becoming a bottleneck. And so we also saw that the junior engineers were able to take on bigger features."
What did you believe that turned out to be wrong?"Assuming that AI would get it by itself was a wrong assumption."
What pushback did you get?"They believe the code base was too complex for AI agent to understand. But when we sort of worked with them and tried to like show other features that we were building, which was all the same complexity in the same code base, it sort of brought them around. And so it was mostly being very supportive and trying to understand why they believe certain thing was not there and then sort of doing a pair AI coding session with them."
Four of the eight answers, trimmed, with one slip of the tongue corrected.

None of that is polished, and it does not need to be. The facts are all there. Now sort them into boxes. Amazon recommends the STAR format (situation, task, action, result) for behavioral answers. We add one more box, the earned secret: the lesson only someone who lived it would know.

Figure 2

The finished story

SituationIn January 2026, we came back from the winter break. I had spent the holidays using a new AI coding model, and it was far better than what we had tried before. My team had tried AI coding without much traction, and a lot of the senior engineers did not believe the code was good enough.
TaskI was an engineer on the team, not its manager, so I could not tell anyone to change. I wanted us to use AI every day so we could deliver more, and that meant bringing the skeptics with me.
ActionI started with proof. I built a feature I was responsible for, got it reviewed and merged, and only then told the team that AI had written every line. They were surprised, and they asked me to present my setup to everyone. We then set up a new workflow on one AI coding tool, and I set up a weekly demo to share what we were learning. I got my manager's alignment to keep the demo going, and asked people to show work in progress, including designs, so we could agree on a design before building it. Some people still believed our code base was too complex for an AI agent. I showed them features of the same complexity in the same code base, and did pair AI coding sessions with them.
ResultEveryone on the team moved to the new workflow. Our PR output went up about 20%, and writing code stopped being the bottleneck. Junior engineers took on bigger features, which gave senior engineers more time for architecture and design.
Earned secretWe had assumed the model would understand our code base by itself, and we were wrong. Even once the model got good, it needed our code structure, our diagrams and plans, and the reasons behind past decisions. The demos taught me something too: they kept going once people could show more than the finished product.
Every fact comes from the session. Read out loud, this takes about two minutes.

What changed between the two, and why

Compare the finished story with what was said. Four changes were on purpose.

  • It keeps one number. The session also produced "10 to 20%" and "it almost felt like it increased team velocity by like 50%". A feeling is not a measurement, and an interviewer can ask how you got it. One of Amazon's Leadership Principles says leaders "are skeptical when metrics and anecdote differ". Keep the number you can defend.
  • "We" became "I" in the action. The session said "we sort of worked with them", and the story says "I showed them". The result can belong to the team. The action has to be what you did, because you are the one being hired, so make this change only where it really was you.
  • The adjectives went. The session said "drastically" or "drastic" four times. The facts do that job better.
  • The resistance stayed. Without the skeptics, this is a project update. Their doubts, and what you did about them, make it a leadership story.

This story still has one thing to fix. "About 20%" needs a comparison: 20% more than when, and over how long? Get that answer before you use it, and be ready to say what those PRs shipped.

Get ready for the follow-ups

Expect the interviewer to start where your story ends. Test yours against these:

  • "What did you do, specifically?" Name the steps only you took. Here: built the feature, set up the demo, paired with the skeptics.
  • "Who disagreed with you, and why?" Give the objection at its strongest, then what changed their mind. Here: they believed the code base was too complex, until they saw work of the same complexity and tried it themselves.
  • "How do you know it worked?" Give the number, what it compares to and how you measured it. With no number, give a named outcome: junior engineers taking on bigger features.
  • "Did it last?" Say what kept it going. Here: the weekly demo, and letting people bring unfinished work to it.
  • "What did you get wrong?" Your earned secret is a ready answer. Here: assuming the model would understand the code base by itself.

Tell the story in about two minutes, then stop and let them ask. The detail you cut is what you use in the follow-ups.

One story also covers more than one question. This one works for "Tell me about a time you influenced without authority", "Tell me about a time you brought skeptical colleagues around" and "Tell me about a change you drove on your team".

Say it out loud before the interview

A story you have only read is not ready. Say it to a friend or to your phone, and listen for the places where you speed up or go vague. Those are the parts you do not know well yet.

You can also do all of this in Koaches, which is where we recorded the session above. Its story coach asks the questions one at a time, by voice or by typing, and follows up on what you say. When you finish, it puts your story into the form, using only what you said. You can then choose to have it rate how strong the story is. Once you have prepped a job, it shows which of that job's questions your story answers, so you can practice them out loud and get feedback on your answer.

Leadership is one of eight must-have stories there, and the same approach works for the others, from conflict to failure.

Build your first story by talking.

Answer the questions, get the write-up, then practice until it comes easily. New accounts get $5 of free credits.

Start preparing

Sources

  1. Google re:Work, A guide to structured interviewing: rework.withgoogle.com
  2. Amazon, What it is like to interview at Amazon: aboutamazon.com
  3. Amazon, Leadership Principles: amazon.jobs