Anduril's 2,090 Reqs vs 85 Cleared Rust Engineers: The Real Math

If you are sourcing for defense tech in 2026, the pool you think you are fishing does not exist. Anduril has 2,090 open reqs, OpenAI and Anthropic stood up cleared FDE teams within days of each other in May 2026, and the visible intersection of "Rust engineer with a clearance-adjacent employer today" is 85 named humans in the entire United States.

Arsenal-1 hit production in March 2026, months ahead of Anduril's July target, and 736 of those 2,090 reqs opened in the last 29 days alone. The role profile - C++/Rust, ROS2, real-time and safety-critical, US citizenship, Secret or TS-eligible - collides head-on with the cleared FDE bidders. Every sourcer chasing this profile is fishing the same pond, and the pond is smaller than anyone in the pitch decks admits.

The pool is 85 people, not thousands

The visible pool of Rust engineers currently employed at US Defense and Space Manufacturing companies is 85. That is the actual intersection Anduril's Arsenal-1 recruiting team is competing over, and it is smaller than most sourcers assume by an order of magnitude.

In Refolk's index of professional profiles, the top employers for those 85 are almost comically concentrated:

• Lockheed Martin: 6
• Anduril: 2
• Saronic Technologies: 2
• Stoke Space: 2
• US Air Force: 2

Six employers account for the entire named surface area. If Anduril hires aggressively from Lockheed, Saronic, and Stoke through Q1 2026, they exhaust the visible defense-Rust pool in a single quarter and then have to invent supply. That is the actual constraint, not comp bands or job-post copy.

85

Why LinkedIn Recruiter shows you an empty screen

A Boolean like (Rust AND C++ AND "TS/SCI") in LinkedIn Recruiter returns essentially nothing, because the cleared-Rust intersection barely exists as a native population. The pool sourcers imagine is not an intersection; it is a union of two adjacent pools that have to be stitched together.

Here is the math behind that:

Only 2.5% of the US Rust population currently works in defense or space manufacturing. The winning motion is not finding the mythical cleared Rust native. It is running two separate searches and converting:

1. Cleared C++ and embedded engineers at Lockheed, Northrop, and SOCOM contractors, then reskilling them into Rust.
2. Rust engineers at Cloudflare, NVIDIA, and Kraken who can pass a clearance investigation, then sponsoring them through it.

LinkedIn Recruiter is structurally bad at the second search because clearance eligibility is not a field, GitHub Rust output is not indexed, and "will pass an SF-86" is a judgment call, not a filter. That is the exact gap Refolk closes: you describe the person you actually need in plain English ("Senior Rust engineer in the US, no visa needed, real-time systems background, has shipped production Rust in the last 18 months") and get a ranked list that draws on GitHub, LinkedIn, and the open web at once.

Anduril's ~$170K median cannot win a comp fight

Anduril's salary band has a median around $170K, and that number cannot win a straight-up bidding war against Palantir's FDE comp, let alone OpenAI or Anthropic. The 2,090 reqs will close on mission, equity, and geography, or they will not close.

The competing bands, per public reporting:

• Palantir FDE average TC: $238K, range $205K to $486K, with staff-level FDEs clearing $630K+.
• OpenAI and Anthropic cleared FDE bands: $350K to $550K for mid-to-senior levels.
• Anduril median: ~$170K.

Recruiters pitching TC lose this fight. Recruiters pitching Arsenal-1 as the only place to ship autonomous weapons at scale win it.

The candidates who take Anduril at $170K are motivated by three things: the Iran munitions burn rate that keeps showing up in Axios reporting, Ohio cost-of-living arbitrage against Bay Area and DC, and equity upside from the $5B Series H at a $61B valuation. Sourcing scripts that lead with base comp are misreading the buyer. Scripts that lead with "150 Fury aircraft per year at full shift capacity, and you own the autonomy stack" convert.

Arsenal-1 is in Ohio. Your ROS2 pool is not.

ROS2 talent geography is wrong for defense, and relocation, not skill scarcity, is the real bottleneck for Arsenal-1's robotics roles. The 3,605 US ROS-skilled engineers in Refolk's index cluster in SF Bay, LA, and Austin, at commercial robotics shops like GrayMatter Robotics, Serve Robotics, Flexiv, and Treeswift.

Arsenal-1 is a 5-million-square-foot campus on 500 acres in Pickaway County, Ohio, with a ~$900M first build-out and a target of 4,000 employees (250 by end of 2026). That is not a "come to the office three days a week" story. It is a full uproot from a warm-weather robotics hub to central Ohio.

The implications for sourcing plans:

• Model ROS2 hires as relocation packages first, comp packages second.
• Weight candidates already outside SF Bay and LA higher, even at equal skill.
• Treat Central Ohio Technical College (COTC), which is spinning up defense-manufacturing courses for Anduril, as a real pipeline for entry-level and technician roles, not a rounding error.
• Do not assume Modern Intelligence and other defense-AI startups will lose their ROS2 hires to Anduril on comp alone. They usually lose on mission fit, not money.

The FDE fight is a Palantir-alumni fight

Every cleared forward-deployed engineer role at OpenAI, Anthropic, and Anduril is fundamentally a raid on Palantir alumni, because Palantir originated the FDE role in 2005 and is still the only shop with a workforce holding active TS/SCI clearances embedded at the Pentagon, FBI, SOCOM, UK MoD, and allied intel services.

The bidding war is not abstract. OpenAI's Deployment Company ($4B+) and Anthropic's $1.5B JV with Blackstone and Goldman Sachs both launched FDE ventures in May 2026, within days of each other. OpenAI signed an agreement with the Department of War committing to cleared forward-deployed engineers embedded with the US government, plus cleared safety and alignment researchers in the loop. Forward-deployed engineer job postings grew more than 800% between January and September 2025.

800%

Sourcing "FDE with clearance" without explicitly naming Palantir alumni in your search is malpractice at this point. The practical moves:

1. Build a Palantir alumni list segmented by team (Gotham vs Foundry vs Apollo) and by tenure bracket. Tenure predicts clearance status better than title.
2. Filter for people who left Palantir between 2019 and 2023. Clearances typically stay reinstatable for two years post-separation without a full reinvestigation.
3. Cross-reference GitHub activity for Rust or modern C++ output. Palantir shops heavily in Java and TypeScript, so recent Rust commits are a strong signal of side-project depth.
4. Treat Saronic (Austin), Stoke Space (Kent, WA), and Anduril's own recent hires as secondary alumni pools. They already re-cleared once.

This is a workflow Refolk is designed for: describe "ex-Palantir engineers in the US who left in the last 3 to 5 years and have shipped Rust or modern C++ on GitHub in the last 12 months" and get the named humans, not a keyword-match dump.

What a realistic Anduril sourcing plan looks like

A realistic plan for filling Anduril-shaped roles in 2026 stops pretending the cleared-Rust intersection is a market and starts treating it as three parallel conversion funnels. If you cannot describe your plan as three funnels, you do not have a plan.

1. Convert cleared C++/embedded engineers into Rust

• Source pools: Lockheed Martin, Northrop Grumman, Raytheon, L3Harris, SOCOM contractors, USAF and USSF civilian engineering.
• Signal to filter on: 5+ years of modern C++ (C++17 or later), any Rust side-project activity on GitHub, real-time or safety-critical shipping history.
• Pitch: Arsenal-1, autonomy stack ownership, equity at $61B valuation.

2. Clear existing Rust natives

• Source pools: Cloudflare, NVIDIA, Kraken, AWS, and the crypto/infra Rust concentration in SF Bay and Seattle.
• Signal: US citizenship, no adverse foreign contacts, willingness to relocate.
• Pitch: mission, plus a clearance sponsorship path measured in months, not years.

3. Raid Palantir and Palantir-adjacent alumni

• Source pools: Palantir alumni 2019 to 2023, plus recent hires at Saronic, Stoke Space, and Modern Intelligence.
• Signal: prior FDE or embed experience, clearance recency, government-customer stomach.
• Pitch: fewer PowerPoints, more shipping.

Founders and defense-tech recruiters who plan around the fantasy pool ("we just need cleared Rust engineers") burn Q1 and Q2 and hit June with a fraction of the plan. The teams that plan around the real pool - 85 native, plus two conversion funnels of a few thousand each - hit their number, because LinkedIn Recruiter's Boolean stack was never going to see the other 2,005 hires anyway.

FAQ

How many cleared Rust engineers are actually available for Anduril to hire?

The visible pool of Rust engineers currently at US Defense and Space Manufacturing employers is 85, per Refolk's index. Six employers (Lockheed Martin, Anduril, Saronic, Stoke Space, US Air Force, and a small tail) account for essentially all of them. Anything beyond that number requires either clearing existing Rust engineers from commercial infra shops or reskilling cleared C++ engineers into Rust, which is why treating the "cleared Rust" search as a single Boolean instead of two conversion funnels fails.

Why does LinkedIn Recruiter miss most of this pool?

Because clearance eligibility is not a LinkedIn field, GitHub Rust output is not indexed, and US citizenship is inferred rather than confirmed. A Boolean like (Rust AND C++ AND "TS/SCI") returns near-zero results, and the actual candidates - cleared C++ engineers with Rust side projects, or Rust natives who could pass an SF-86 - sit in two separate searches that Recruiter cannot stitch together. Sourcing tools that read GitHub and the open web alongside LinkedIn recover most of the missing surface area.

Can Anduril actually compete with OpenAI and Anthropic on comp for FDEs?

No, not on straight TC. Anduril's median is ~$170K against Palantir's $238K FDE average and OpenAI/Anthropic's $350K to $550K bands. Anduril's winning pitch is mission (Arsenal-1, Iran munitions burn rate, Barracuda ramping to 5,000 units per year by end-2026), Ohio cost-of-living arbitrage, and equity upside from the $5B Series H at a $61B valuation. Recruiters leading with base salary lose these candidates; recruiters leading with what the candidate ships win them.

What is the single biggest sourcing mistake defense-tech recruiters make in 2026?

Treating "cleared FDE" as a search instead of a named alumni raid on Palantir. Palantir originated the FDE role in 2005 and is still the only shop with a workforce holding active TS/SCI clearances embedded at the Pentagon, FBI, SOCOM, and allied intel services. OpenAI, Anthropic, and Anduril are all fishing the same few thousand ex-Palantir engineers, and any sourcing plan that does not explicitly name Palantir alumni as its primary source pool is going to lose those candidates to the labs on comp before it ever gets a first call.