FDE Sourcing Broke in 18 Months: 3,382 Profiles, 4 Titles, 1 Playbook

If you are still pasting "Forward Deployed Engineer" into LinkedIn as your primary string, you are working a minority of the actual pool. The title has fragmented into at least four public variants in 18 months, and the ex-Palantir lineage that seeded the role has almost entirely moved on to companies that no longer use it.

Perspective AI's June 2026 sweep of roughly 1,000 live FDE job posts confirmed the fragmentation. Meanwhile, OpenAI's Omer Khan, who co-created the Forward Deployed AI Engineer role at Palantir, is publicly hiring across five cities on two continents. This piece uses Refolk's index numbers to map the pool, name the non-obvious hunting grounds, and lay out a sourcing playbook that stops leaning on the title string.

Why "Forward Deployed Engineer" is now a minority signal

The canonical title covers roughly 72% of the actual talent market, which means a title-only search silently discards about 28% of qualified candidates before the first outreach. Refolk's index shows 3,382 US profiles currently holding one of the four fragmented title variants, and in a sampled top-titles cluster only 18 of 25 still use the exact string "Forward Deployed Engineer."

The mechanism is boring but decisive: title standardization lags role proliferation by 12 to 18 months, and applied-AI startups deliberately rebrand to differentiate from Palantir's operational lineage. The four variants currently competing for the same job are:

• Forward Deployed Engineer (the Palantir canonical)
• Forward Deployed AI Engineer (Khan and Yannis Karamanlakis coined this at Palantir roughly two years before its OpenAI reboot)
• Applied AI Engineer (frontier-lab preferred)
• Deployment Solutions Engineer (enterprise and cloud preferred)

1,000%+

FDE hiring grew more than 1,000% year over year through early 2026 on Perspective AI's count. When a job category grows that fast, the naming convention breaks first. If you are still writing forward deployed engineer sourcing strings against a single title, you are optimizing for the 2024 market.

The ex-Palantir "Delta" lineage is the real lever

Sourcing on employer-history "Palantir" plus a skill cluster beats sourcing on title, because roughly 95% of the original Delta lineage has already left. Refolk's index shows 378 US profiles with a Palantir/FDE-adjacent pedigree signal. Only 20 are still at Palantir Technologies. The rest have diffused into applied-AI startups, frontier labs, and founding teams, and most of them dropped "Forward Deployed" from their current title.

The word "Delta" itself is worth adding to your queries. Palantir's earliest on-site FDEs were called Deltas, and until 2016 Palantir had more FDEs than software engineers, an unusual ratio that shows how central the embedded model was from the start. A LinkedIn keyword search on "Delta" plus "Palantir" surfaces the lineage more reliably than any current title string.

This is the exact gap Refolk closes: instead of stacking booleans, you describe the profile in plain English ("ex-Palantir engineers, now at applied-AI startups, 4+ years of customer-facing deployment work") and get a ranked shortlist across GitHub, LinkedIn, and the open web. Ex-Palantir engineer sourcing is a pedigree-plus-shape problem, not a title problem.

Where the Deltas landed

Refolk's index flags a specific cluster of employers absorbing the diffused Delta pool. Most boolean strings never surface them together:

• Palantir (the residual 20)
• Northslope Technologies
• Gecko Robotics
• Modal
• Amp
• Roboflow
• Cresta

None of these are obvious FDE hunting grounds if you are pattern-matching on "frontier lab." All of them are shipping the customer-embedded model under a different name.

The numbers, in one table

Here is the dataset the rest of this playbook rests on. All figures are US and drawn from Refolk's index unless noted.

The two numbers to internalize: 3,382 and 378. The first is your addressable pool across the four titles. The second is your highest-leverage subset, and 94.7% of it is already outside Palantir's walls.

Why FDEs surface under sales-adjacent titles

The role is 47% customer-facing on median, which means the best candidates often sit under Solutions Engineer, Deployment Strategist, or founding-team titles rather than IC-engineer titles. FDEs spend a median 47% of their week customer-facing, 31% shipping code, and 22% on internal coordination, per Perspective AI's 2026 survey of 1,500 FDEs. That work profile does not read as "engineer" on a profile summary.

Practical consequence: if your sourcing filter excludes sales-adjacent titles, you are excluding the actual talent. Titles worth explicitly including:

1. Solutions Engineer / Solutions Architect (especially post-sales)
2. Deployment Strategist
3. Customer Engineer
4. Founding Engineer at applied-AI seed startups
5. Technical Account Manager at Palantir, Databricks, or Snowflake

This is where forward deployed AI engineer recruiting stops being a keyword game. The Perspective AI data on time allocation tells you the ideal candidate reads like a hybrid of a Palantir FDE and an enterprise Solutions Engineer, which no single boolean captures.

Geography is bimodal, not SF-dominated

New York leads Refolk's US FDE sample, not San Francisco, and Khan's OpenAI hiring spans five cities on two continents. In the sampled cluster, 7 of 25 profiles sit in New York and 4 in San Francisco, with Seattle, Boston, and Austin filling out the top five. If your search is Bay-Area-anchored, you are missing the NYC frontier-lab cluster where teams are actually growing.

Khan's public LinkedIn recruiting in August 2026 covers SF, NYC, Paris, London, and Dublin/Munich for OpenAI's Life Sciences vertical. That is a five-city, single-vertical push that is invisible to any US-only, single-title search. When the hiring manager who co-created the role is running a five-city net, matching that geographic breadth is table stakes.

The recruiters winning FDE searches in 2026 source the way Omer Khan recruits: by network, city cluster, and problem-shape, not title string.

Cities worth adding to your applied AI engineer hiring searches, based on Khan's public activity and Refolk's index sample:

• New York (28% of sampled US pool)
• San Francisco (16% of sampled US pool)
• Seattle, Boston, Austin (the long tail that matters)
• London, Paris, Dublin, Munich (Khan's European net)

Compensation is bifurcating, and the stealth pool sits in the gap

Frontier-lab FDE comp now runs 2.0 to 3.5x Palantir's classic FDSE at the same level, which creates a predictable outreach window: Palantir Deltas 6 to 12 months from a vesting cliff. Median mid-level FDE total comp in 2026 is $385K, staff-level is $610K, and principal FDEs at frontier labs clear $1.2M, per Perspective AI's 2026 compensation report on 1,200 FDEs. Equity is now 55 to 70% of comp at the top, up from 35 to 45% in 2024.

The stealth pool has three properties:

• Currently at Palantir or an enterprise tier company
• 6 to 12 months from a vesting cliff or refresh event
• Not actively posting or updating profile summaries

That pool is invisible to job-board scraping and to inbound. It is only reachable if you can query pedigree, tenure, and current employer together, which is a natural fit for a plain-English query in Refolk rather than a stack of LinkedIn filters. The comp math alone will move a meaningful fraction of that 20-person residual Palantir Delta group when someone actually reaches them.

The playbook, in six moves

Stop treating FDE as a title search. Treat it as a pedigree, shape, and city search. Two demand-side facts frame why this matters now: AWS announced a $1 billion investment in a dedicated Forward Deployed Engineering organization in June 2026 to embed thousands of engineers with customers under Solutions Architect-adjacent naming, and OpenAI's careers page listed 40+ FDE-matching roles globally as of August 4, 2026. Both are invisible to a single-title search.

3,382

1. Query all four variants together: Forward Deployed Engineer, Forward Deployed AI Engineer, Applied AI Engineer, Deployment Solutions Engineer. Add Delta as a keyword.
2. Layer employer-history Palantir on top. That single filter unlocks the 378-person lineage pool, 95% of which no longer holds the title.
3. Include sales-adjacent titles: Solutions Engineer, Deployment Strategist, Customer Engineer, TAM at Palantir/Databricks/Snowflake.
4. Widen geography beyond SF. NYC leads the US sample; add Seattle, Boston, Austin, London, Paris, Dublin, Munich.
5. Prioritize outreach to candidates 6 to 12 months from a Palantir vesting event. The 2.0 to 3.5x frontier-lab premium does the persuasion work.
6. Hunt inside Northslope, Gecko Robotics, Modal, Amp, Roboflow, and Cresta before hunting inside OpenAI and Anthropic. The diffusion is already there.

If you want a shortcut through steps 1 to 4, Refolk turns the whole thing into one natural-language query and returns a shortlist across GitHub, LinkedIn, and the open web in a single pass. That is the actual answer to how to hire FDE in a market where the title is a lagging indicator.

FAQ

What are the four FDE title variants I should search?

Forward Deployed Engineer, Forward Deployed AI Engineer, Applied AI Engineer, and Deployment Solutions Engineer. Perspective AI's June 2026 analysis of ~1,000 live posts confirmed all four are competing for effectively the same role. Add "Delta" as a keyword to catch the original Palantir lineage, and include sales-adjacent titles like Solutions Engineer and Deployment Strategist because FDEs spend a median 47% of their week customer-facing.

How big is the ex-Palantir FDE pool, really?

Refolk's index shows 378 US profiles with a Palantir/FDE-adjacent pedigree signal, and only 20 are still at Palantir Technologies. That means roughly 94.7% of the Delta lineage has already diffused into applied-AI startups, frontier labs, and founding teams. The residual 20 are worth a targeted outreach around vesting cliffs, but the leverage is in the 358 who left.

Why is New York now the leading US city for FDE talent?

Frontier labs are building their customer-embedded teams east, not west. In Refolk's sampled US pool, 28% of profiles sit in NYC versus 16% in SF, and OpenAI's Omer Khan is publicly hiring across SF, NYC, Paris, London, and Dublin/Munich as of August 2026. The FDE role follows the customer, and OpenAI's current push is anchored on the Life Sciences vertical announced by Oliver Jay at Fortune Brainstorm AI 2025.

Is boolean sourcing dead for FDE roles?

Boolean sourcing is not dead, but title-only boolean is now a filter that discards the majority. Roughly 28% of the live FDE pool uses a non-canonical title, and about 95% of the Palantir Delta lineage has changed employer since the role was coined. The winning approach is pedigree plus shape plus city, expressed in plain English rather than nested AND/OR strings, which is why natural-language sourcing tools produce shortlists that title booleans cannot reach.