DIU's FPV Pilot Registry Is Locked. The Signal Isn't: 946 vs 16.

On Oct 8, 2026, the Defense Innovation Unit confirmed it is recruiting roughly 100 FPV drone pilots for Gauntlet III of the Drone Dominance program, and building an uncapped standing registry of operators across every service. The registry is the sourcing database every defense-tech company would pay to query. It is gated behind a .mil email address, and you cannot have it.

You can mirror the signal, though. And once you look at what the open talent market actually contains, you will realize mirroring it is the only way to hire for this archetype at all.

Why the registry matters even if you cannot touch it

The registry matters because it is the first public admission that FPV operator talent is a national-scarcity problem, not a credential problem. DIU program test lead Erin Cherry put it in one line: "There's no cap on the registry. We want to hear from any warfighter with FPV experience."

The policy stack behind that sentence is doing the real work:

• A June 6, 2025 executive order directed DoD to buy and train with low-cost American drones.
• A July 10, 2025 Pete Hegseth memo reclassified small drones as consumable munitions and set a goal that every squad fields them by end of 2026.
• Ukraine is producing 6 to 7 million FPV drones a year. The Pentagon is trying to close that gap on the airframe side and, implicitly, on the operator side.

Squad-level drone saturation implies a squad-level pilot curve. The US has never trained operators at that density before, and the private side (Anduril, Shield AI, Neros, Skydio, Red Cat) is competing with DoD for the same bodies. Cherry's registry is the first centralized attempt to count them.

Defense-tech recruiters are structurally shut out: DIU will only reply to applicants on a valid .mil address, and participation routes through the services. So the question is not "how do I get into the registry." It is "how do I build my own index of the same people, scoped to the ones who have already left or are about to."

What the open market actually contains: 946 vs 16

The open market contains almost no self-identified FPV pilots, and one dominant employer of record for everyone else. In Refolk's index of professional profiles, there are 946 US profiles with titles containing "UAS Operator," "Drone Operator," "UAV Pilot," or "Unmanned Aircraft Systems Operator." There are 16 with "FPV drone pilot" or "FPV drone operator" in the headline. That is a 59x gap.

Two things fall out of that table. First, the US Army is effectively the only large employer of record for drone operators in the open index, holding 14 of the top 25 slots. That is the mechanical reason DIU can build a cross-service registry in the first place, and the reason Anduril, Shield AI, Neros and Red Cat are all fishing in a pond of transitioning soldiers numbered in the hundreds, not the thousands.

Second, FPV as a resume term barely exists. Zero hits inside Defense & Space industry employers. The people you want are not writing "FPV" on their LinkedIn. They are writing "11C," "19D," "UAS Operator," "RQ-7 Shadow," "150U," or nothing at all because they are still in uniform.

59x

The operator-engineer archetype nobody is advertising for

The archetype every serious drone company is actually hiring is an operator-engineer: someone who has flown FPV in a tactical setting and can also instrument a test flight, write a test card, or ship firmware. It is one job pretending to be two, and the open-market signal for it is almost zero.

Look at how the current postings describe it:

• Neros Technologies is hiring an FPV Flight Instructor in El Segundo that reports into a Mobile Training Team. The company produces about 1,200 airframes a week and raised a $250M Series C at a $2.5B post-money in August 2026, co-led by Sequoia. The instructor role did not exist as a startup job family 18 months ago.
• Shield AI is hiring Forward Deployed Engineers and autonomy flight-test engineers in San Diego, with explicit preference language for "prior DoD programs, military UAV operations" and FAA Part 107.
• Anduril has passed 1,500 employees and is posting Integration Engineer and Manufacturing Engineer roles in Costa Mesa at $129K to $194K bands.
• Red Cat / Teal Drones won the Army SRR program of record (Black Widow), which makes their pilot-facing engineers a public proxy for FPV operator demand.

Three of those four will reject a Part 107 civilian pilot who has never flown a quad through a window at 60 mph. Part 107 is an FAA line-of-sight commercial certification. FPV combat flying is a different motor skill with a different failure mode. Treating them as interchangeable is the single most common sourcing mistake in the sector right now.

Part 107 is a driver's license. The registry is a demolition-derby roster. Stop pretending they are the same filter.

Mirror DIU's taxonomy, not LinkedIn's

The playbook is to mirror DIU's taxonomy inside your own ATS and sourcing workflow, because the taxonomy itself is the differentiator. DIU is making three choices worth copying verbatim.

1. Recency over seniority

Cherry's "five years or five months" line is deliberate. FPV skill decays in months because the airframes, radios, and goggles change every cycle. A recruiter's "10+ years UAS experience" filter actively excludes the best candidates, because the best candidates were in high school when the current meta was invented. Replace tenure filters with a recency filter: flew FPV in the last 90 days, in any context.

2. Any service, any MOS

The registry is cross-service on purpose. The talent is distributed across Army 11C mortarmen and 19D cavalry scouts, Marine Raider teams, Air Force Special Tactics, Navy EOD, and National Guard units that have rotated through Camp Grayling, MI, where Gauntlet II staged 49 companies and roughly 79 drones. If your Boolean searches "Army UAS" only, you miss two thirds of the pool.

3. Volume at the top of funnel, screen later

The registry is uncapped. Yours should be too. The mistake is pre-qualifying for ITAR, clearance, and skill in the first search. Build the biggest possible pool, then filter.

This is the exact gap Refolk closes for defense-tech sourcing: you describe the operator-engineer you want in plain English (something like "US Army soldier who flew FPV at Grayling in the last two years and has any software or RF background") and get a ranked shortlist across GitHub, LinkedIn, and the open web. Title Boolean cannot do this. The research above is the reason: the titles you would search for are not on the resumes you want.

The ITAR gate narrows a narrow pool further

ITAR eligibility collapses the realistic sourcing universe to US persons, which overlaps almost perfectly with the registry you cannot see. Neros's job postings require candidates to be a "U.S. Person" or otherwise eligible for an export license. That language is standard across Anduril, Shield AI, Red Cat and AeroVironment.

The practical consequence: the global LinkedIn pool for "drone engineer" is largely irrelevant to this sector. The relevant pool is US-citizen, veteran-heavy, and already clustered around a handful of bases and training events. If you are sourcing from the EU or India for FPV-adjacent roles, you are wasting cycles that your competition at Anduril and Shield AI is not.

Named geographies that matter more than cities

• Camp Grayling, MI - Gauntlet II venue, 49 companies tested roughly 79 drones.
• Torrance and El Segundo, CA - Neros footprint, Mobile Training Team origin.
• San Diego - Shield AI autonomy and flight-test engineering.
• Costa Mesa - Anduril integration and manufacturing.

Gauntlet rankings as a free scouting feed

Treat the public Gauntlet leaderboard as a scouting feed, because DoD published it for you. Neros placed second in Gauntlet I, first in close-quarters battle in Gauntlet II, and third in deep strike in Gauntlet II. That means Neros's flight-test engineers and FPV instructors are, right now, the most poachable identified-by-name talent in the sector.

The mechanism is simple: ranked competition inside a classified-adjacent program creates a public performance signal that no interview panel can replicate. If you are hiring a flight-test lead, the top finishers in every Gauntlet category are your shortlist, whether or not those engineers are "looking." Red Cat's Black Widow win on the Army SRR program is the same signal in a different wrapper.

Each public program outcome is a named cohort of engineers whose fingerprints are on the delivery. Lifting those cohorts out by employer, timeline, and role is faster than reading press releases and faster than paying a generic sourcing tool to run the same Boolean worse.

1,200

The 90-day playbook

The 90-day playbook is to mirror DIU's registry inside your own pipeline before Gauntlet III closes, because the registry is pulling the same people you want toward DoD programs, not toward you. Do it in this order:

1. Rewrite your Boolean. Drop "10+ years" and "Part 107 required." Add MOS codes (11C, 19D, 150U, 15W), base names (Grayling, Benning, Bragg, Pendleton), and program names (SRR, Drone Dominance).
2. Build the ITAR-eligible pool first, then filter for skill. Scope to US persons. You will lose fewer finalists at offer stage.
3. Index the Gauntlet leaderboards. Pull the named engineers behind the top-three finishers in each category, every cycle.
4. Stop waiting for the resume keyword. FPV is not on resumes. Use plain-English search that indexes GitHub commits, open-web profiles, and LinkedIn together, so a soldier with one line about "small UAS at Grayling" and three Rust repos surfaces as one record.
5. Open a dedicated instructor track. Neros already has one. If you do not, your flight-test engineers will burn out training new hires instead of flying.

The registry will fill. Gauntlet III will run. The 100 pilots DIU names will become the next 100 hires at Anduril, Shield AI, Neros, Skydio and Red Cat, in some order. The companies that mirrored the taxonomy early will get the top of that distribution. The companies still filtering on Part 107 will get the bottom.

FAQ

Can defense-tech recruiters access the DIU FPV pilot registry directly?

No. DIU officials will only respond to Pilot Interest Form submissions from a valid .mil email address, and all pilot participation routes through the individual military services. There is no commercial API, no partner tier, and no plausible workaround. The practical move is to mirror the registry's taxonomy (any service, any experience level, recency over seniority) inside your own sourcing stack, and to pipeline transitioning operators before they separate.

Why is the gap between generic UAS operators and FPV pilots so large?

Because FPV is a motor skill and a tactical practice, not an FAA credential, so it does not land on resumes. In Refolk's index, 946 US profiles claim UAS/UAV/Drone Operator titles but only 16 claim FPV. The people flying FPV at scale are active-duty infantry and special-operations soldiers whose public profiles, if they exist, show their MOS and unit rather than the airframe they fly. Title-based Boolean will miss them every time.

Which companies are competing hardest for this archetype right now?

Anduril (1,500+ employees, aggressive Costa Mesa hiring), Shield AI (Forward Deployed Engineers and autonomy flight-test in San Diego), Neros (1,200 airframes a week out of El Segundo, Mobile Training Team), Skydio ($825M raised), and Red Cat / Teal (Army SRR program of record winner with Black Widow). AeroVironment is also expanding tactical and loitering-munitions programs.

What is the single highest-leverage change to make this week?

Rewrite the "required experience" line in every FPV-adjacent job description to match Cherry's "five years or five months" standard, and strip Part 107 from the requirements block unless the role is strictly civilian. That one edit unlocks the pool of recently separated junior enlisted operators who are the actual target population and who currently self-select out of your funnel at the first bullet point.