You can’t hire forward deployed engineers… so make your own.

Let's start with the honest truth. There is no supply of FDE’s. Or nowhere near enough. Not of the people who can actually drive AI transformation inside an organization. Yes the models are here and getting better every month but what's scarce is the humans who turn them into outcomes.

Now the market is finally saying out loud what a lot of us already knew. Christian & Timbers estimates there are only about 2k engineers in the US with the specific cocktail of sector knowledge, gravitas, and hands-on applied-AI experience needed to consistently deliver enterprise ROI.

This is out of roughly 17k who carry the title at all.

LinkedIn found this job category grew 42x between 2023 and 2025. More than three times faster than just “AI engineer.” Meanwhile the qualified candidate pool only grew roughly 50% against postings that grew 800%.

Which sounds solvable enough of a problem… until you go looking to hire these people.

We all love labeling things but frankly this label is a little overused.

What you're really looking for is a world class operator who can go from a customer conversation, internal or external, to a product built, iterated, and deployed, and then work with every stakeholder to drive adoption and own the outcome. That's it.

If you strip away the buzzword, it’s really these four core tenets:

The term was coined at Palantir in the early 2010s (“Deltas”/ stole from an old war term where you staffed units on the front lines to learn vs the safety of camp) and has since been picked up by OpenAI, Anthropic, and Google. The titles are now fragmenting across 6+ variants because nobody can agree on what to call it.

What it actually is is someone who can own discovery, build, and production rollout. Their work is measured by adoption and workflow impact and not by hours or even headcount.

I actually had Claude ingest a dozen job posts and the gist of it is that, at its core, this is “a research role wearing an engineering badge, with a consultant's touch.”

Traditionally the software engineer was the obvious pick. Build it, deploy it, work with stakeholders to get adoption. With AI though, you can now build a ton of different products without a formally trained software engineer. That single fact changed the archetypes that work.

Now there isn't just one profile but at least three.

The archetypes tell you which pools to fish in to start. The attributes below tell you who's actually going to be great. These are just some of the ones people don't talk about enough:

We covered some archetypes and attributes but it’s just the start. There’s a reason why finding all this is hard. You can’t power search your way in Linkedin to get the answer…

You can't just type “forward deployed engineer” into a database, pull up a million people, fire off cold outreach, and call it done. That talent pool doesn't exist as a searchable list. These people are labeled with a dozen different titles, and they're all still climbing their own personal AI learning curve. Same as the rest of the world.

The market confirms it. LinkedIn title search is the wrong tool for this role. The people who source it well look for deployment signals instead like real infrastructure code, shipped products, a failed startup someone actually ran. They also look for customer-facing receipts such as public writing that translates hard technical problems for non-technical readers, and field-CTO or deployment stints.

So the way to get this right is two moves:

That's a pretty different muscle than keyword matching. It's pool identification + human, relationship-driven + qualitative assessment. Which, not coincidentally, is exactly the kind of search this moment demands. And exactly where a network-and-assessment model beats a job board.

We think referrals and referencing is the only way to validate this…