Shield AI's 300 New Reqs Meet a 95-Person Autonomy Bench

Shield AI opened 300+ new reqs in the last 30 days and sits at 755 open jobs as of October 4, 2026. The exact-title US bench for the role it is hiring hardest, "Autonomy Engineer," is 95 people. If you are sourcing against this surge with a title filter and a "cleared" Boolean, you are fishing a pond you can count on two hands.

Shield AI's hiring surge, in numbers

Shield AI has 755 open jobs with 300+ added in the last 30 days, roughly 60% in engineering, after a $1.5B Series G in March 2026 at a $12.7B post-money valuation. Headcount is near 1,888. The round was led by Advent International with JPMorganChase's Security and Resiliency Initiative co-leading, and Blackstone providing $500M in fixed-return preferred equity. The hiring is downstream of the Air Force Collaborative Combat Aircraft program, which drove a 140% valuation jump from $5.3B twelve months earlier.

The practical read: Shield AI is now a top-three destination for US autonomy talent and is pulling from the same bench as Anduril (June 2025, $2.5B Series G at $60B), Skydio, and Applied Intuition. Nobody is backfilling the pond.

0.13

The sourceable pool is 95, not thousands

Only 95 US profiles in Refolk's index currently hold the title "Autonomy Engineer." That is the entire national candidate universe for the exact role Shield AI is backfilling 300 times this month. The adjacent "Robotics Engineer / Robotics Software Engineer" pool is 2,188, so there is a 23x reframing lever available, but only if you stop recruiting against a job title and start recruiting against a skillset.

Here is the shape of the market in one table.

The derived number worth staring at is 0.13. Shield AI alone has less than one existing US autonomy engineer per open seat, and that is before Anduril, Skydio, Boeing, or Lockheed opens a single competing req. The title filter is the trap: a Boolean on "Autonomy Engineer" returns a spreadsheet, not a pipeline.

Where the 95 actually work

The top employers of US autonomy engineers are a 10-company list: Skydio, May Mobility, Pratt Miller, Applied Intuition, Anduril, Boeing, Zipline, EpiSci, Parrot, and Monarch. If you are not messaging into those ten, you are not sourcing autonomy; you are hoping. Skydio and May Mobility each hold 4 of the top 25 in Refolk's index, Pratt Miller holds 3, and the geography concentrates in the San Francisco Bay Area, Ann Arbor, and Pittsburgh.

The mechanism matters. Autonomy is a tight apprenticeship discipline - state estimation, controls, motion planning, perception stacks integrated under real-time constraints - and people who do it learned it at a place that lets them do it on real hardware. Reframing the req around the underlying skills (C++ on real-time Linux, control theory, sim-to-real) is the only honest way to expand supply.

That reframing is the exact gap Refolk closes for defense recruiters: you describe the person in plain English and get a ranked shortlist across GitHub, LinkedIn, and the open web, not a title match.

Clearance is a lagging constraint, not a leading one

Clearance filters do not create candidates; they only subtract them, and the subtraction is too slow to keep pace with Shield AI's req velocity. DCSA's median for a Tier 5 Top Secret investigation is 150 to 200 days, so sponsorship-based hiring literally cannot scale to 300 reqs a month. The only fast lever is poaching already-cleared engineers from the primes and the cleared-adjacent startups.

A few numbers to frame how tight that is:

• 67% of defense and aerospace firms report they cannot fill open cleared roles (AIA Workforce Report).
• More than 70,000 security-cleared positions remain unfilled across the industry (jobswithdod.com).
• AI/ML engineers in the cleared space are among the "nearly impossible to fill" categories (Federal News Network, Jan 2026).

The practical implication for a Shield AI recruiter: treat clearance as a lane-selector, not a screener.

• Lane one: active TS/SCI holders at Lockheed, Northrop, Raytheon, L3Harris, and Boeing Defense, who can start in 30 days.
• Lane two: Secret-eligible engineers at Anduril, Skydio, and Applied Intuition, who can be sponsored up.
• Lane three: clearable commercial talent (US persons, clean background) sponsored from zero, 6 to 7 months to seat.

Any req like R6023 (Senior C++ Engineer, Foundations) or R6035 (Senior Software Engineer, Autonomy Integration) that cannot articulate which lane it is in will sit for a quarter.

70,000+

The comp story has flipped

Shield AI's $185,000 median advertised salary across 236 postings sits above many prime contractor bands, which closes the historical "big-tech pays more" gap for cleared talent. The remaining friction is not cash; it is mission fit, program secrecy, and relocation to San Diego.

That reframes the pitch. For a prime-contractor engineer sitting in Huntsville or El Segundo, Shield AI is now a comp upgrade on top of a mission upgrade. For a commercial autonomy engineer at Applied Intuition or Zipline, it is a mission trade at roughly flat cash, with the real negotiation being equity in a $12.7B company that just priced a 140% markup. Sourcers who still open with "we can match your base" are using a 2022 script.

Three concrete adjustments based on the comp flip:

1. Lead with program, not pay. The CCA selection is public and specific; name it.
2. Treat equity as the headline for commercial poaches. The March 2026 round is a dated, nameable anchor.
3. For prime-contractor poaches, lead with velocity and keep the $185K median in the second message.

The non-engineering 40% is a separate playbook

About 300 of Shield AI's 755 open reqs are non-engineering - capture managers, program managers, manufacturing, operations - and sourcing them from an SF autonomy startup bench is the wrong map. Reqs like R6187 (Strategic Capture Manager, USG Data Agencies, active Secret required, TS preferred) compete against Lockheed, Northrop, and Booz Allen talent pools in the DC metro, not against anybody in the Bay Area.

Two different sourcing muscles inside one company:

• Engineering (60%, ~455 reqs): San Diego, San Francisco Bay Area, Dallas. Poach from Skydio, Anduril, Applied Intuition, EpiSci. Skills first, title second.
• Capture and program (40%, ~300 reqs): Washington DC metro, Huntsville. Poach from Lockheed, Northrop, Raytheon, Booz Allen. Clearance first, program experience second.

Teams that try to run one playbook across both halves will underperform both halves.

Shield AI's hiring is not a volume problem. It is a 10-company map problem, and 300 reqs a month is going to force every recruiter to learn it.

The Ukraine channel most recruiters ignore

Shield AI is actively posting Drone Operator and Senior Autonomy Software roles in Kyiv and Lviv, and almost no US-based defense recruiter is sourcing against those nodes. Ukrainian autonomy engineers have shipped more real-world drone autonomy under adversarial conditions than any comparable Western cohort since 2023.

For a US recruiter, the Kyiv and Lviv postings are useful two ways:

• As a direct hire path for engineers already permitted to work out of Shield AI's European entities.
• As a signal map. Engineers who worked at Ukrainian autonomy firms, or at Western firms with Kyiv offices, between 2023 and 2026 are an under-indexed pool of relocation-willing candidates for the San Diego HQ.

The signal lives in GitHub commits, conference talks, and open-web mentions, not in clean LinkedIn titles. "Engineers who contributed to open-source UAV autonomy stacks or wrote about counter-UAS work from Kyiv or Lviv since 2023" is a Refolk query. It is not a LinkedIn query.

The 10-company poaching map, in order

Start where title concentration is densest, then widen. Based on Refolk's top-employers list for US autonomy engineers:

1. Skydio and May Mobility. Highest title density (4 each in top 25). Commercial comp, no clearance baseline, so expect a 6 to 7 month runway on sponsorship.
2. Pratt Miller. 3 in top 25, defense-adjacent, better clearance baseline than pure commercial.
3. Anduril. Direct peer, often cleared already. Hardest to poach but highest fit; expect counter-offers.
4. Applied Intuition and EpiSci. Sim and autonomy stack experience; EpiSci is the cleared-adjacent half.
5. Zipline, Boeing, Parrot, Monarch. Round out the top 10, each with 1 to 2 names in the usable bench.

Run the first message against clusters 1 and 2 for raw volume, and clusters 3 and 4 for cleared conversions. Clusters outside these ten exist, but they are not where the apprenticeship happened, so expect a lower response-to-pass-screen rate.

FAQ

How many of Shield AI's 755 open roles are engineering?

Roughly 60%, or about 455 reqs. The top skill tags across 236 postings are Python (181 mentions), C++ (146), Linux (96), SolidWorks (57), and Docker and Kubernetes (30 each). On the physical-AI subset, Simulation (50), Real-Time Systems (37), Embedded Systems (34), and Control Theory (17) index much higher than they would at a commercial AI company, which is the strongest signal of who the ideal hire looks like.

What is the realistic time to hire for a cleared autonomy role at Shield AI?

30 days for an already-cleared poach from a prime or cleared startup, 60 to 90 days for a Secret-eligible candidate who needs sponsorship, and 150 to 200 days for a clearable commercial engineer starting from zero (DCSA Tier 5 median). Shield AI's req pace of 300 per month means at least two of those three lanes have to be running in parallel at all times, with the 30-day lane doing most of the work.

Should I source outside the US?

Yes, specifically Kyiv and Lviv, where Shield AI is actively posting Drone Operator and Senior Autonomy Software roles. Ukrainian autonomy engineers have shipped more adversarial-conditions UAV work than any Western cohort between 2023 and 2026, and most US defense recruiters are not touching the pool. Use GitHub-first sourcing rather than LinkedIn, because the signal is in commits and conference talks, not in titles.

Who is Shield AI competing against for the same 95 people?

Primarily Anduril (which raised a $2.5B Series G in June 2025 at a $60B valuation), Skydio, Applied Intuition, Zipline, and the autonomy groups inside Boeing and Lockheed. The 10-company top-employers list in Refolk's index is both Shield AI's poaching map and its competitor set, which is why the market will not clear through pay alone and why mission, program, and equity are the three levers that actually move candidates.