Founder-engineer fit

For the first three to five engineers you hire, the thing that matters most after raw skill is not culture fit. It is whether the engineer fits you.

Culture fit assumes a culture. At hire one, there isn't one.

Almost all hiring advice assumes a company already exists. There are norms, a way code gets reviewed, a way disagreements get settled. The advice tells you to hire people who fit that.

At hire one there is nothing to fit. There is you, maybe a co-founder, a half-built product, and a set of habits you have not written down because you did not know they were habits.

The famous research is about a different stage. Google's Project Aristotle looked at 180 teams with a median of nine people and found psychological safety mattered most.³ True, and useless for a two-person company. Sam Altman has said that whatever the founders do becomes the culture.² Peter Thiel says every company is a culture.⁴ Both are saying the same thing: early on, the culture is just you.

So fit comes in layers. Founder fit first. Team fit once there is a team. Culture fit once there is a culture. Founders burn their first hires by screening for the third layer before the first one exists.

What founder-engineer fit is

Three things.

How they think. Judgment under uncertainty, appetite for risk, how they argue. Do they reach for the same kind of answer you would, or at least an answer you can reason with?

How they communicate. Can they write clearly. Do they flag problems before the problems are big. Does a conversation with them add clarity or remove it.

How they work. Pace. Attention to detail. Whether they can direct themselves when there is no process to lean on.

Being pleasant is not fit. Being a strong engineer is not fit. Both are required, and neither is enough.

The closest analogy is co-founder matching. Tempo, how each of you handles being wrong, what happens when the plan stops working. Those are the same questions, asked of someone with less equity.

There is research behind this. A review of 172 studies found that fit with your direct boss is one of the strongest predictors of job satisfaction.⁵ Gallup puts at least 70 percent of the variance in team engagement on the manager.⁶ At a three-person startup, the manager is the founder. There is nobody else.

Fit is tested at the edges of a task, not in the middle

JD Pruett ran seven hiring experiments and wrote them up in October 2026. The framing that stuck with me is 20/10: what matters is what it is like to work with someone in the first 20 percent of a project and the last 10 percent.⁷

The first 20 percent is framing. What are we building, how should it be shaped, what is the taste of the thing. The last 10 percent is getting it into real users' hands and dealing with everything that breaks on the way.

The middle 70 percent is execution. Models are eating that part. It is also the only part a classic technical interview measures.

So fit shows up at the ends. And as the middle shrinks, the ends become a bigger share of the job, not a smaller one.

What the founders who got it right did for their first ten

John Collison says no later batch of ten people will shape the company as much as the first ten.⁸

Brian Chesky spent four or five months and talked to hundreds of people before hiring Nick Grandy as Airbnb's first engineer. One of his questions: if you had ten years to live, would you still take this job?¹

The Collisons worked a week with most of their first ten before making an offer.⁸

Mark Zuckerberg only hires a direct report if he would work for that person.⁹

Jeff Bezos, in his 1998 shareholder letter, asked three questions: will you admire this person, will they raise the average of the group, and along what dimension might they be a superstar?¹⁰

Ben Horowitz: hire for strength, not for the absence of weakness.¹¹

Paul Graham's phrase for the ideal founder is "relentlessly resourceful."¹² It is also the phrase for the ideal second engineer.

None of these are culture-fit tests. They are all personal.

The first engineers are a specific type of person

The engineer who does great work at a 200-person company is not the one who does great work as employee number two. The second type is rarer.

Big companies have docs, conventions, QA, on-call rotations, someone who already decided how the repo is laid out. Early on, the engineer is building that system while shipping inside it.

George Mack calls it high agency: clear thinking plus bias to action plus a willingness to be disagreeable.¹³ PostHog just says "cracked."¹⁴

This is not a ranking of engineers. Most great engineers belong at a later stage, and the company will need them. The test for hire two is simple. Hand them a job with no process around it and watch. Some people light up. Some tense up.

Mistakes are cheap early and get expensive later

Early, a mistake costs almost nothing and teaches a lot. Later, a mistake costs a lot and teaches little. The two curves cross somewhere around product-market fit.

That matters for hiring because an early engineer makes dozens of small product decisions a week. There is no PM. They have to be comfortable using their own judgment and comfortable being wrong in public.

The best early engineers build the smallest version, ship it, and look. The ones who struggle produce diagrams and design reviews before anything exists. I am not arguing for sloppiness. I am arguing that the fastest way to learn what to build is to build a little of it.

My own fit profile, as a worked example

This is personal, so here is mine.

Pace. If a task is small, I want it done now, not queued. I once paired with a very methodical engineer who worked a queue in order. He was good. We were both miserable.

Try and see. My default is to build the smallest version and observe. I need engineers who share that default, or who can switch into it when I ask.

Call me out. I want to be disagreed with early, with reasons, and then I want us to commit. Over-agreeable engineers let my bad ideas ship.

Details. There is no QA team. The engineer's own testing is the testing. I want people who click through the product the way a user would before they tell me it is done.

At TeamEx I have hired capable engineers who did not match this. They were great engineers. They were not my engineers. That distinction is the whole essay.

Three founders, three different fits

Three founders I have worked with, three different briefs.

Clipboard Health wanted high-agency, curious engineers who build before being asked.

Maple wanted strong communicators, especially in writing.

TopHelpers wanted generalists with one area of real depth.

The same engineer could be the perfect hire for one and a miss for the other two. That is what it means for fit to be personal.

The vibe match, and the other common mistakes

Vibe match without a technical bar. Fit is a tiebreaker, not a substitute. "We had a great chat" is usually a sign the interview had no structure.

Interviewing only the middle 70 percent. LeetCode rounds measure execution, which is the part AI absorbs first.

Hiring the resume. A big-name employer tells you the person passed that company's bar. It tells you nothing about how they work when there is no bar.

Outsourcing the judgment. An agency can verify skills. It cannot tell you whether someone matches your tempo. Thiel's line is that recruiting should never be outsourced, and he is right about the early hires.⁴

How I would run the search today

1. Write the brief in plain English. Something like: an AI engineer who has built LLM products at a pre-seed startup, ships fast, tests their own work, and pushes back when I am wrong.
2. Set the specialty and expertise. AI engineer; LLM applications, agents, evals.
3. Add skills. Python, TypeScript, LangGraph, vector databases.
4. Add products they have built that look like yours. A recruiting platform, a two-sided marketplace.
5. Pick the strengths you actually need. High agency, bias to action, comfort with ambiguity, clear communication, ownership.
6. Pick the stages they have worked at. Pre-seed, seed.
7. Add your own criteria in plain sentences. Three lines about the person you want to work with.
8. Run it and treat the result as a shortlist. It finds people you would enjoy working with. The interview still decides.

When founder-engineer fit stops being the only test

Around the fifth engineer, team fit becomes something you can test. Past fifteen, culture fit becomes real.

Founder-led hiring usually runs through the first 50 to 100 people, with the founder keeping the final interview. But the intense, personal kind of fit I am describing is a first-five thing. Get those right and the later layers have something to grow on.