What is Talent and Talent Engineering

Traditional recruiting has three roles.

The recruiter owns a pipeline. They review inbound applications. They run the recruiter phone screen. They do cold outbound and move people through stages, fill the role, close the req, move to the next one.

Recruiting ops keeps the recruiter organized. They own the ATS, the data, the scheduling, the reporting.

HR deals with the people who already work at the company. Onboarding, benefits, performance, conflict, compliance.

This model has worked for decades. It is built for volume and for predictability.

But it was built for a world where the job was to fill roles. That is not the game the top 1% of operators are playing anymore.

The game changed because the stakes changed.

At a high talent density company, one hire is not one hire. It sets the bar for the next ten. Stripe gave away more than 10% of the company to its first ten people because John Collison's view was that you aren't hiring ten people, you're hiring a hundred. The first ten bring the next ten.

When each hire is that consequential, "owning a pipeline" is the wrong frame. You are not processing applicants. You are identifying a small number of exceptional people, building real relationships with them, and convincing them to bet their career on you.

That is a different job. It needs a different kind of person. So we got a new word.

A "talent" person is not a fancier recruiter. The word exists because the job stopped being about pipelines.

Here is my definition.

Talent = anyone who is great at attracting smart people and has deep relationships with all of them.

They do not need to be technical. They do not need a recruiting background at all.

Usually they are extroverts. They love talking to people. They love helping people.

You find this person across the whole ecosystem now. At VC firms, where the talent partner is often the most connected person in the building. At startups, where they are employee number five and somehow know everyone worth knowing. At frontier labs, where the entire recruiting strategy is one person's network and reputation.

A great talent person is a magnet. People want to work with them, so people want to work where they are.

Talent engineering is the other new role, and it is genuinely different.

Talent engineering = an engineer who loves to build people tools and is also a great engineer.

I want to be heavy about the second half, because it is the part people skip. A talent engineer is not a recruiter who learned to code. They are not an ops person who can write a script. They are a real engineer. The kind who could be a member of technical staff, who could be shipping the core product, who would clear the bar at any top company. They just happen to love building people tools instead.

That is the ideal. A great engineer, good enough to build the product, who points that talent at hiring because they actually love it.

And that is exactly why it is so rare. Anyone good enough to be a talent engineer is good enough to do something that looks more prestigious and pays in more obvious ways. The pull is always toward the product. So the person who is both genuinely great at engineering and genuinely drawn to people tooling is one of the hardest profiles to find in the whole market. When you find one, you hold onto them.

There are other definitions floating around, and some companies draw the line differently. This is how I define it.

The cleanest way to understand both roles is to notice how much they overlap.

A great talent person can be a great engineer.

A great talent person can also just vibe code their own tool, or use something off the shelf like Juicebox, and get most of the leverage without writing real software.

And a great engineer can build the tools and do the recruiting. We have seen exactly this at Cognition. They are referral only, and they have an engineer building tools for the talent team as a side project.

So the Venn diagram has a lot of middle. The labels matter less than the underlying truth, which is that the best people on both sides are builders. One builds relationships. One builds tools. The strongest operators do both.

The a16z Talent Engineer Fellowship describes a talent engineer as "an engineer who recruits, not just a recruiter who codes." I agree with that completely. It is close to my own definition, and it is the right instinct. Where we part ways is on what you do with it.

Their sharper framing is that recruiting is not an assessment problem, it is a many-to-many matchmaking problem, the right person in the right role at the right moment, and matchmaking is an engineering problem.

The first half is right. The second half is where I get off.

You can engineer who you should match. You can build the sourcing agents, the market maps, the relationship graphs, the referral systems. That layer is real and it is valuable, and the best teams invest in it heavily. But you cannot engineer whether the match takes. You cannot engineer whether someone trusts you. The identify and engage layer is an engineering problem. The close layer, the relationship layer, is a human one. Confusing the two is how you end up with a beautiful pipeline and no one who actually wants to work with you.

The market does not understand the role and who to hire. Most companies read the archetype and go hire a recruiter who can write a little code, or an ops person who is good with tools. That is not the same person, and it is not close.

The perfect scenario is a great engineer who is willing to do recruiting full time. Not a recruiter who learned to code, but a real engineer, someone who could be a member of technical staff, who chooses to point all of that at hiring. They bring the technical depth that makes a candidate genuinely excited, the kind of conversation where someone walks away thinking these are my people and this is the hardest, most interesting problem I could be working on.

But that person is extremely hard to find. Anyone good enough to be a great engineer is good enough to do something that looks more prestigious and pays in more obvious ways, and the pull is always toward the product. So you cannot count on finding one. This is exactly why culture matters so much. If you cannot guarantee one great engineer doing recruiting full time, the next best thing is a culture where every engineer treats recruiting as part of the job, where posting a name in the hiring channel is normal, where spending a quarter of your week chasing talent is a natural habit. Culture is what turns the whole team into recruiters when you do not have the one rare person who would do it alone.

The other version I see working, and it is pretty common at some of the highest talent density organizations, is a great talent person paired with a deeply technical engineer. The talent person brings the best top of funnel, is deeply respected, and has real relationships with the candidates. They are the reason the right people pick up the phone at all. The engineer brings the technical depth that makes the candidate excited once they do. One opens the door. The other makes them want to walk through it. Neither could do it alone, and together they are very hard to beat.

Because at the end of the day, the best recruiters are still the founders and the engineers themselves. The tools help. The systems help. But the thing that actually makes someone exceptional want to come is another exceptional person they respect, telling them about a problem worth their life. You cannot engineer that. You can only build a company full of people worth saying yes to.

And in the end the title does not really matter. Talent, talent engineer, recruiter, whatever you want to put on the door. But I will say this plainly: you are not an actual engineer unless you are a member of technical staff. Personally, I would not trust technical feedback on a candidate unless it came from a member of technical staff. The people who can give real technical signal and make a real technical pitch are the ones who have actually done the work, and that is exactly why the founders and the engineers will always be the best recruiters in the building.

If you look at how the highest talent density companies hire, almost none of them run the traditional model. They run some blend of talent and talent engineering instead.

Start with Cursor. Adam Ward, who runs talent at Cursor and built the firm behind hiring at OpenAI, Anthropic, Stripe, Notion, and Figma before they acquired his team, describes most recruiting as a funnel of doom. You reach out to a hundred people, twenty reply, and you hire the best of those twenty. But the twenty who replied are not the top twenty percent. They are whoever you happened to catch on a bad day. So you assess out the rest and end up with the best of a pretty good lot. Do that for years and you get an above average company, never an exceptional one.

Cursor's answer is to treat every role like an executive search. You scope it, you market map who the genuine top one percent are, you form a hypothesis that these specific people are the best, and then you use your process to validate that hypothesis instead of using it to weed out bozos. The atomic unit is a person, not a job spec. Most companies start with a hole in the org, open a req, source a list, and fill it. Cursor starts with a name. Someone really good gets posted in a hiring channel in Slack, the team swarms them with attention, and if the interest is mutual they start on Monday. "Looking for a job" is not a requirement for getting an offer there.

The sourcing is not LinkedIn title searches. Two people on the team spend fifty to sixty hours a week doing nothing but mining referrals from current employees, by hand. Not pulling a list of names and moving on, but sitting down with someone, digging through their network, and triangulating which team at which company at which moment was actually great, then chasing everyone who was on it. They found one engineer because he was running Cursor workshops in Stockholm. They found another because he showed up as an outlier in their own usage data, coding with Cursor late into the night. When a talent pool opens up, like a company shutting down or a round of layoffs, the whole team goes looking for the best people in it. Anyone who sees an impressive tweet, release, or blog post drops the name in the channel with "should we hire?"

Then everyone closes, not just a recruiter. They gave an early Stripe and Notion designer, an Apple fanboy, a vintage Macintosh. They flew to Germany twice for one engineer who kept saying no, because as they put it, no is often the start of the conversation. They hand delivered an espresso machine to a candidate's new house, sourced through her own interior designer, after six months of casual coffee chats. Members of technical staff spend around a quarter of their time recruiting. Leadership signs off on every hire.

Ward's framing for why this works is the part most people miss. No one cares more about a hire than the hiring manager. The moment recruiting becomes recruiting's job, the moment effort gets handed to the function, it is almost irrecoverable. The best companies never hand it off. The most important thing the company has is talent, so the people in the building do the work themselves.

Cognition runs a different version of the same idea. They are referral only. Their founding team's competitive programming network is the sourcing engine. And they have an engineer building tools for the talent team, purely out of passion. That is talent and talent engineering sitting in the same company, doing exactly what each is best at.

But the thing that defines Cognition is how far the founders will personally go to close. Scott Wu tells a story about a candidate who was a junior at MIT, very young, who interviewed better than almost any full time candidate they had ever talked to. They asked him to take time off school and come build Devin. He wanted to, but his parents wanted him to graduate, and he could not see how to make it work. So the founders flew to North Carolina, went straight from the airport to his parents' house, had dinner with the family, and worked out a setup where he could work essentially full time and still come in for the classes he needed to graduate. Everyone ended up happy. Then they flew straight home.

Scale's Alexandr Wang has said recruiting is the thing he spends the plurality of his time on, and that he either personally interviews every hire or signs off on every candidate packet. His framing is that a recruiting team that looks like a college admissions office is certain death for a startup. Done right, recruiting looks less like sifting applicants and more like courtship, where the spark sometimes comes after months of pursuit. Stripe's founders gave away more than ten percent of the company to their first ten people, and worked alongside most of them for a week before making an offer. Anduril runs a drone racing contest with a job attached, because they would rather watch you build than read your resume. Ramp hires for slope over pedigree and fills the company with people other firms misprice. Notion treats its user community as a hiring pool and keeps Ivan Zhao in nearly every loop.

At xAI we were running the combination. Some of our best youth hires came through young members of technical staff and the co-founders themselves, not through a recruiting funnel. We had an MTS mapping out the tooling and also helping close, bringing people in personally. And we leaned hard on events. We held several where we simply asked our MTS to invite their friends. The engineers were the magnet and the closers at the same time. That is the whole model in one company: talent and technical depth working the same candidate, with the best people in the building doing the recruiting themselves.

None of these are pipeline companies. They are network companies with good tools. Recruiting is not a department they delegate. It is something the best people in the building do themselves.

Here is the part most people get wrong. They try to judge a talent person by their resume, their pipeline numbers, or their process. None of that tells you anything.

You judge a talent person by the people around them.

Three things matter.

They are respected by extraordinary people. The most talented person you know should light up when their name comes up. And the people inside the company should feel the same way. When someone is close with people internally and respected within their friend circle, mining referrals becomes very easy.

Their friend group is exceptional. Look at who they actually spend time with. The magnet and the people stuck to it are the same signal. Great people cluster, and a real talent person is always at the center of a great cluster.

And they genuinely love being around smart people. They would rather have dinner with someone brilliant than do almost anything else.

That is the whole test. It sounds too simple, but it isn't. If extraordinary people respect them, if their circle is excellent, and if they light up around smart people, they will be good at this job. If any of the three is missing, no amount of recruiting experience will fix it.

The reason this works is that the job is recursive. A talent person's entire value is that exceptional people want to be near them. So the best evidence that they can attract exceptional people is that they already have.

Talent is a person who attracts smart people and holds deep relationships with them. They may or may not be technical. They are the magnet.

Talent engineering is an engineer who builds people tools, usually because they love it. They are the leverage.

The old model split the work into recruiter, recruiting ops, and HR, and optimized for filling seats. The new model collapses it into magnets and builders, and optimizes for landing a small number of exceptional people.

You can engineer how you find and reach people. You cannot engineer whether they want to come. The companies that understand which half is which are the ones winning the talent game.