Booz Allen Just Released 3,300 Cleared Engineers. LinkedIn Returns 9.
Booz Allen's $922M civil collapse pushed thousands of TS/SCI engineers into the market. Here's how to source them before Anduril and Palantir do.
On May 22, 2026, Booz Allen Hamilton printed FY26 revenue of $11.2 billion, down 6.4% year over year, and disclosed that customer-facing headcount had fallen from 32,700 to 28,800. That is roughly 3,900 client-facing employees gone in twelve months, the back half of a contraction CEO Horacio Rozanski kicked off in May 2025 with a 2,500-person cut "heavily concentrated" in the civil division. If you recruit for Anduril, Palantir, Shield AI, or any commercial AI vendor with a federal arm, this is the largest involuntary release of cleared engineers in recent memory, and most of them are not on LinkedIn by design.
The math nobody is doing out loud
Booz Allen's March 2025 investor presentation laid out the pre-cut profile in unusual detail: 22,000 technologists, 6,000 top-tier software engineers, 8,000 cyber professionals, 2,500 AI practitioners, and 70% of the firm holding security clearances. Apply that 70% cleared rate to the 3,900-person customer-facing reduction and you get a hard floor of 2,700 to 3,300 cleared employees pushed into the market across FY26.
The trigger was civil-specific. Civil revenue fell from $4.17B in FY25 to $3.25B in FY26, a roughly $922M drop, while Defense ($6.07B) and Intelligence ($1.90B grew modestly. The expiration of a single $1.56B VA contract accounted for 13% of 2025 civil revenue. Five other major civil contract run rates were reduced in April 2025. The cohort exiting is not random: it skews toward engineers who built AI, cloud, and data platforms for IRS, VA, HHS, and CMS.
That is the most useful framing for a commercial AI recruiter: these are not "defense engineers." They are regulated-deployment engineers. They have shipped production systems under FedRAMP High, ATO timelines, and audit regimes that look a lot like what Anthropic, OpenAI, and Scale need for their federal arms. The market still prices them as defense-only, which is the arbitrage.
Why LinkedIn returns single digits
Run a literal Boolean for "TS/SCI" against U.S. software engineer profiles on any professional network and the result is embarrassing. Our own index of the open professional web turned up 9 matches against that exact string. Nine. Against a federal cleared population that ODNI pegged at 4.3 million as of October 2015.
This is not a bug in your Boolean. It is OPSEC working as designed. As ClearanceJobs has put it for years, "secret squirrels are difficult to spot and nearly impossible to hire, because they'd rather not be found online." Cleared engineers are trained, repeatedly, not to list compartments, program names, customer agencies, or in many cases even the clearance level itself on public profiles. The signal you are searching for has been deliberately scrubbed at the source.
LinkedIn Recruiter's keyword index makes this worse, not better. It rewards explicit self-labeling. The cleared population systematically does not self-label. So every defense-tech recruiter using the same "TS/SCI" + "Polygraph" + "active clearance" Boolean is fishing the same shallow pool of people who either forgot the OPSEC briefing or just left the cleared world entirely.
The signal you are searching for has been deliberately scrubbed at the source. That is not a Boolean problem. That is a sourcing model problem.
What actually works
Stop searching for the clearance. Start searching for the pattern that implies it.
Employer history. Booz Allen, Leidos, SAIC, CACI, ManTech, Northrop Grumman IT, Lockheed Martin IT, Peraton, Raytheon BBN. Engineers with five or more years stitched across two or more of these firms are cleared with probability approaching one.
Geography. Cleared engineers cluster by base proximity, not by tech hub. Fort Meade pulls Columbia and Hanover, Maryland. NRO and NGA pull Chantilly and Herndon, Virginia. Space Force and NORTHCOM pull Colorado Springs. NGA-West and Buckley pull Aurora, Colorado. Our index data shows ex-civil engineers concentrating in Fairfax, Colorado Springs, and Aurora, with Second Front Systems showing up as a frequent destination.
Open-source breadcrumbs. Federal agency GitHub orgs (18F, USDS, CMS, VA) have public repos with public contributor histories. Cross-reference contributors against Booz Allen email patterns in commit logs. DEF CON, Shmoocon, and BSides DC speaker rosters. SANS instructor pages. These are the leakage points.
The redeployment clock
Here is the part most recruiters miss. Rozanski has been explicit that the cuts produced "a significant number of employees needing to be redeployed simultaneously." That is corporate code for: Booz Allen is trying to move these people onto Defense and Intel contracts internally before they walk. If they get re-onboarded onto a TS/SCI program at Booz, Leidos, or SAIC, they are effectively off-market for 12 to 24 months.
The May 2025 cohort moved through their 90-day separation through summer 2025. The senior-rank cuts driving Q4 FY26 hit the market right now, summer and fall 2026. After that, the surviving names go back into the cleared-prime ecosystem and disappear behind a SCIF door.
This is the specific sourcing problem Refolk was built for. You describe the person in plain English (employer tenure, base-proximate ZIP, federal repo contributions, security conference talks) and get back a ranked shortlist drawn from GitHub, LinkedIn, and the open web together. The clearance keyword does not need to appear anywhere in the query, because it does not appear on the profile.
Where the cohort is actually landing
The named destinations in the data are instructive.
Anduril is the loudest buyer. The company is now valued near $30 billion and is standing up Arsenal-1, a roughly 5 million square foot manufacturing campus south of Columbus, Ohio, that needs more than 4,000 people. Much of that headcount requires clearance. Palmer Luckey's team ran an "AI Grand Prix" autonomous drone competition in early 2026 with a $500,000 prize pool, where the winner skips standard recruiting and goes straight to hiring-manager interviews. That is not a gimmick. It is an explicit acknowledgment that the normal cleared-recruiting funnel is broken and they need a side door.
Palantir continues to pull from the same pool with FDE comp at senior bands that can clear $450K total, competitive with anything a big prime offers when you weight the equity. The Palantir Meritocracy Fellowship pulls in pre-clearance high school graduates and sponsors them through, which is the long game. The short game is hiring the civil-division engineer who already has the badge.
Shield AI is aggressive on forward-deployed autonomy talent. Second Front Systems keeps surfacing in our index as a destination for ex-Booz civil engineers, which makes sense given its focus on getting commercial software accredited for DoD environments. Anthropic, OpenAI for Government, and Scale's federal arm are all hiring into roles that look a lot like FDE plus FedRAMP plus mission delivery, which is exactly what civil engineers spent the last decade doing for IRS and VA.
A playbook for the next 120 days
If you are recruiting against this window, here is the order of operations.
1. Build the employer-history net first. Ex-Booz civil engineers with 3+ years tenure, Senior Consultant rank or above, separated between January and December 2026. Cross-reference with Leidos, SAIC, CACI alumni who also touched VA, IRS, HHS, or CMS programs.
2. Layer geography. Filter to ZIPs within 25 miles of Fort Meade, Fort Belvoir, Chantilly, Quantico, Colorado Springs, Aurora, Huntsville, San Antonio, Tampa. The cleared-adjacent population concentrates here whether or not the profile says so.
3. Add the GitHub pass. Pull contributor histories from federal agency GitHub orgs and the major civil-program open source repos. Match names against your employer-history list. This is where Refolk earns its keep: you can ask for "engineers with commits to USDS or 18F repos who also list a Booz Allen or Leidos role" and get a single ranked output instead of running three searches and reconciling spreadsheets.
4. Work the niche boards, but don't expect an edge. ClearanceJobs, ClearedConnections, and ClearedJobs.net are necessary because the cleared community uses them. They are not sufficient because every defense-AI recruiter is already there. The edge is the off-platform list you built in steps 1 through 3.
5. Lead with the regulated-deployment pitch, not the defense pitch. A civil engineer who spent six years getting VA systems through ATO does not want to hear "come build weapons." They want to hear that their FedRAMP High and audit experience is the rare credential that Anthropic federal or OpenAI for Government will pay a premium for. The framing matters.
6. Move now. The May 2025 cohort is already redeploying. The Q4 FY26 senior cuts are mid-flight. By Q1 2027, the survivors are back inside SCIFs and the window closes.
The structural fact to internalize: cleared sourcing is not a keyword problem, it is an inference problem. The platforms that index self-reported labels will keep returning nine results. The recruiters who win this cycle are the ones who stopped asking the profile to identify itself and started asking the pattern to.
FAQ
How many cleared engineers did Booz Allen actually release in FY26?
The disclosed customer-facing reduction was approximately 3,900 employees between March 2025 and March 2026. Applying Booz Allen's own disclosed 70% clearance rate gives a floor of 2,700 to 3,300 cleared exits, heavily weighted toward the civil division where revenue fell roughly $922M year over year. That is the largest single-firm release of cleared technical talent in recent memory.
Why doesn't LinkedIn surface these candidates?
OPSEC norms in the cleared community discourage listing TS/SCI status, compartments, customer agencies, or program names on public profiles. LinkedIn's index rewards explicit self-labeling, which the population systematically avoids. A literal "TS/SCI" search across U.S. software engineer profiles in our index returned 9 results, which illustrates the gap. Sourcing has to shift from keyword matching to inference based on employer history, base-proximate geography, and federal open-source contributions.
Where are ex-Booz civil engineers actually going?
The visible destinations are Anduril (which needs more than 4,000 people for Arsenal-1 in Columbus, Ohio), Palantir, Shield AI, Second Front Systems, and the federal arms of commercial AI vendors including Anthropic, OpenAI, and Scale. A large share are also being internally redeployed onto Defense and Intelligence contracts at Booz Allen itself, or laterally to Leidos and SAIC, which is why the timing window matters.
What is the realistic window to source this cohort?
The May 2025 layoff cohort moved through 90-day separations starting in summer 2025 and many are already placed. The senior-rank cuts that drove the FY26 Q4 numbers are hitting the market in summer and fall 2026. Engineers re-onboarded onto cleared contracts at primes are effectively off-market for 12 to 24 months, so the actionable window for direct outreach closes by roughly Q1 2027.