LinkedIn Shows 136K FDE Jobs. The Real Pool Is 1,190.

On August 3, 2026, June AI came out of stealth with a $20M pre-seed led by Marc Benioff's Time Ventures, pitching a product that replaces the forward deployed engineer. That same week, LinkedIn kept happily returning "136,000+" matches for "Forward Deployed Engineer" while the actual global population of people who hold an FDE-family title is roughly 7,130, and the Palantir-descended slice most founders actually mean is 1,190. If you are trying to hire one of these people right now, that gap is your entire problem.

Why LinkedIn shows 136,000 FDE jobs when the real title pool is 7,130

LinkedIn's "136,000+" is a Boolean-match artifact, not demand. The search matches any posting containing "engineer" plus "deployed" plus adjacent tokens, so it sweeps in DevOps SREs, field engineers, solutions engineers, and every ATS crosspost that mentions deploying anything to production.

The real ceiling is smaller than the job count by more than an order of magnitude. The entire population of humans who currently hold an FDE-family title, at any seniority in any country, is about 7,130 per Refolk's index. That is the total addressable universe. Every sourcer chasing the 136K number is boiling an ocean whose fish do not exist.

114x

The mechanism is worth naming, because it also breaks Indeed, Wellfound, and most ATS-forwarded aggregators: keyword systems cannot distinguish "Forward Deployed Engineer" (a specific role Palantir originated circa 2006) from "engineer who was deployed forward" (nonsense generated by any two adjacent tokens). Semantic search that understands the role as a concept, not a string, is the only fix. That is the exact gap Refolk closes for this search: describe the role in plain English ("Palantir-descended FDE with 3+ years at one shop, US, ships production Python") and get a ranked shortlist instead of 136,000 field-service reqs.

The real Forward Deployed Engineer talent pool, in numbers

Roughly 7,130 people worldwide hold an FDE-family title today, 3,156 of them in the US, and only about 1,190 have any Palantir orbit in their history. That last number, not the LinkedIn one, is the market you are actually competing in.

Here is the shape of the pool, from Refolk's index cross-referenced with public postings:

Two things jump out. First, in a 25-profile sample of Palantir-context FDEs, 23 still listed Palantir as their current employer, with next-largest employers at n=1 each (RANGR Data, Fourth Age). The diaspora has barely started. Second, the US concentration is absurd for a national title: NYC (7), SF (5), Austin (2), and a small DC-area cluster account for most non-Palantir demand. London is the largest ex-US hub.

If you are running forward deployed engineer sourcing against that shape and still filtering by title only, you are competing with every other founder for the same handful of obvious names.

Why FDE recruiting in 2026 is a graph problem, not a keyword problem

Sourcing an FDE is a Palantir-alumni graph problem, not a title-search problem. The pool is small enough and dense enough that second-degree edges through Palantir, Anduril, Anthropic, and OpenAI's applied teams outperform any keyword filter you can write.

Three mechanics drive this:

1. The alumni network is loyal and clustered. Palantir alumni have founded 111+ companies raising $11.6B+, including Anduril. Until around 2016, Palantir deployed more FDEs than core software engineers. That is a decade-plus of a shared vocabulary, a shared interview format, and shared war stories about the duct-taping era Fast Company described. One warm intro through that graph consistently opens several more.
2. The category name is fragmenting on purpose. Anthropic labels the role "Applied AI Engineering." OpenAI still calls it "Forward Deployed Engineering." Both require production engineering plus customer-facing range. Founders searching only "FDE" miss half the real applied AI engineer talent pool.
3. The winning Boolean is narrow. The single highest-signal public query is ("Applied AI" OR "Forward Deployed" OR "Deployment Strategist") AND (Palantir OR Anduril). Everything else is noise. Even that Boolean caps out around 1,190 profiles globally.

The 136K LinkedIn number describes ATS spam. The 1,190 Palantir-orbit number describes your actual competition.

Where the real roles actually live

Palantir is still the single largest employer of the title, with 95 open FDE-family reqs as of July 9, 2026 (61 in the US), and posted base bands of $135K to $200K with total comp $171K to $295K. Outside Palantir, the current tier-2 US FDE employers surfacing in the index are a tight list, and it is the list to watch for outbound reply rates and competitive intel on comp:

• Cognition
• Modal
• Amp Code
• Roboflow
• Cresta
• Gecko Robotics
• Northslope Technologies
• Hadrius
• Wonderful

None of these will show up cleanly on a LinkedIn title search. Several use "Applied AI Engineer" or "Solutions Engineer" as the public title and reserve "FDE" for internal use. This is the point at which most keyword pipelines silently fail.

What June AI's $20M actually means for the FDE market

June AI's thesis, that AI can replace FDEs, is bullish for FDE comp in the short term, not bearish. Someone has to build the automation, and the only people who can are the ex-FDEs who understand the failure modes of hand-tuning agents against enterprise stacks.

June's round was led by Time Ventures with Michael Dell, Aaron Levie, George Kurtz, Diane Greene, SV Angel, Conviction Embed, Abstract, A*, and Vesey Ventures backing. The founding team (Efrat Rapoport, Ohad Hen, Barak Goldstein, Idan Tsitiat, ex-Bonobo AI and Salesforce) framed it clearly. Rapoport's public line: "AI, paradoxically, increases the demand for professional services." Her longer pitch, that enterprises are spinning up FDE teams to hand-tune agents against Salesforce, ServiceNow, Databricks and Workday and "that approach scales headcount rather than software," is the exact market June is trying to compress. Named early customer CMG, a US mortgage lender, committed at Salesforce's conference to run 100 agents and then stalled in meetings with FDEs and consultants, per CSO Paul Akinmade. That is the wedge.

Every FDE-replacement startup will hire ex-FDEs to build it. That, plus Anthropic, OpenAI, and Databricks paying AI FDE packages that rival staff-level engineers at Meta and Google, is what keeps the pool small despite years of hiring: the people who leave one FDE shop get bid up by three others before their notice period ends.

How to actually source Forward Deployed Engineers in August 2026

Stop searching by title. Start with the ~1,190-person Palantir-orbit graph, filter for candidates who have stayed 3+ years at one FDE shop, and expect to compete on comp with Anthropic and OpenAI.

Here is the playbook to run this quarter:

1. Anchor on the graph, not the keyword. Start from a seed list of 30 to 50 Palantir alumni in your target city, then walk second-degree edges through Anduril, Credal (founded by Jack Fischer, ex-Palantir 2017-2022), Cognition, and Modal. This is how hiring forward deployed engineers actually works in 2026, not a Boolean on a job board.
2. Filter for tenure, not activity. Bob McGrew, ex-Palantir head of product and ex-OpenAI CRO, warns founders copying the model: "don't. The whole thing is a trap that's dangerously easy to turn into a low-margin consulting firm." Most companies advertising FDE roles will fail to retain them. Candidates who have stayed 3+ years at one FDE shop survived a full deployment cycle. The eight-shops-in-six-years profile is a churn signal.
3. Search all three names for the role. Query "Applied AI Engineer," "Forward Deployed Engineer," and "Deployment Strategist" as one pool. Anthropic and OpenAI split it between them on title alone.
4. Price to the real band. Palantir's posted FDE base is $135K to $200K, total comp $171K to $295K. Anthropic and OpenAI pay above that. If your offer opens under Palantir's midpoint for a senior IC, you are not competing.
5. Use interview format as a filter, not a hazing ritual. Palantir's case-style FDE interview rejects roughly 99% of applicants. Copying the format is fine. Copying the reject rate when your pipeline is 40 candidates deep, not 4,000, burns a quarter.
6. Watch the 111 alumni-founded companies. Any of them going through a down round or a layoff puts a small, high-signal batch of ex-Palantir FDEs on the market for 30 to 60 days. Set alerts.

The one insight most founders miss about the FDE pool

The pool stays around 1,200 Palantir-orbit profiles not because nobody is hiring, but because churn is structural. FDE work looks like consulting when the product is not ready, and most FDE employers never get the product ready. Fast Company's framing of Palantir as "technology-powered consulting" for most of its history, with FDEs and deployment strategists doing the duct-taping, is exactly the trap McGrew is warning about.

The corollary for recruiters: the ~1,190 number is not a stock, it is a flow. People rotate in from adjacent solutions-engineering roles, spend a couple of years getting good, then either get bid up to a staff-equivalent applied AI seat or leave the category entirely for founding roles. That is why Palantir alumni have founded 111+ companies at $11.6B raised, and why the diaspora keeps refilling itself from within.

It is also why Palantir's AIP Bootcamp matters as a filter. The five-day pilot on the customer's own data reportedly converts about 75% of prospects and compresses a 12- to 18-month sales cycle to weeks, which is what makes the FDE a revenue-generating unit rather than a cost center. Candidates who look like FDEs on paper but have never shipped that kind of pilot are not who you want. The people who have shipped one, and stayed, are. And on a public pool of 1,190, the fastest way to isolate them is a semantic query, not a Boolean.

FAQ

How many Forward Deployed Engineers actually exist worldwide in 2026?

Roughly 7,130 people hold an FDE-family title globally, with about 3,156 in the US. The subset with Palantir orbit in their background, which is the pool most founders actually mean when they say "FDE," is closer to 1,190. Against a LinkedIn keyword count of 136,000+, the mismatch is why forward deployed engineer sourcing feels impossible on a job board and tractable on a graph.

Why does LinkedIn return 136,000 results for "Forward Deployed Engineer"?

LinkedIn's search matches any post containing overlapping tokens, so DevOps SRE reqs, field-engineer postings, and generic "deployed to production" descriptions pollute the count. Treat the LinkedIn number as a proxy for keyword spam, not demand.

Where do Palantir FDE alumni actually go after Palantir?

They cluster tightly. Destinations include Anduril, Anthropic, OpenAI's applied teams, and a long tail of alumni-founded companies (111+ raising $11.6B+ to date, per public data). Current tier-2 US employers of the title in the Refolk index include Cognition, Modal, Amp Code, Roboflow, Cresta, Gecko Robotics, Northslope Technologies, Hadrius, and Wonderful. Geo-wise, NYC and SF dominate, with a smaller Austin and DC-area cluster, and London leading outside the US.

Does June AI's $20M round mean FDE hiring is about to slow down?

The opposite, in the short term. June AI's own pitch, that enterprises hand-tuning agents against Salesforce, ServiceNow, Databricks, and Workday "scales headcount rather than software," presupposes a large FDE labor market to disrupt. Every startup betting on FDE-replacement will need ex-FDEs to build the replacement, which is why Anthropic, OpenAI, and Databricks already pay AI FDE packages that rival staff-level engineers at Meta and Google.