Anduril's factories can retool in days. The reason is a software stack it built from scratch — and is still hiring to finish.

Inside ArsenalOS: The Software Backbone Behind Anduril's Factory Ambitions

Anduril Industries built ArsenalOS as a proprietary digital-manufacturing platform to collapse the gap between engineering design and factory-floor execution. The system connects every stage of production, from an engineer's initial blueprint through to the technician assembling hardware, within a single integrated environment. Where legacy defense manufacturing fragments that workflow across separate tools for design, quality management, procurement, and assembly, ArsenalOS governs the entire chain from start to finish, Anduril's own description on its website reports.

The platform functions as what Anduril calls a "software-defined manufacturing" system, using enterprise resource management to drive each step of the design, development, and mass-production process, as detailed by Mobility Engineering Tech. That framing matters: it positions the software layer not as a support tool wrapped around hardware production, but as the primary system through which production is planned, tracked, and controlled.

Anduril has staked its two most significant factory investments on ArsenalOS as their operational core. The company's Arsenal-1 facility in Pickaway County, Ohio, a nearly $1 billion mega-factory that began assembling autonomous combat drones in March 2026, months ahead of its original July target, runs on the platform as its foundational software layer, integrating design, development, and mass production for all Anduril products built there, per Anduril's own materials on the Ohio project. Ohio Tech News and JobsOhio report the facility is projected to create roughly 4,000 direct jobs and drive an estimated $2 billion in regional economic impact.

The same software backbone extends to Anduril's Quonset Point facility in Rhode Island, where the company produces autonomous underwater vehicles. The geographic spread (Ohio for aerial combat drones, Rhode Island for maritime systems) means ArsenalOS has to serve as a unified production stack across product lines that share little in hardware but need identical rigor in traceability, quality control, and throughput management.

The hiring data reflects how central the platform is to Anduril's manufacturing thesis. Zero G Talent's board shows Anduril added 160 roles in the past seven days alone, spanning manufacturing test engineers in Quonset Point, data engineers for quality intelligence in Costa Mesa, and NDT inspectors and metrology engineers in Santa Ana. These aren't generic factory hires; they're roles that sit at the intersection of production hardware and the software systems orchestrating it. ArsenalOS isn't a side project at Anduril. It's the operating system for the company's entire manufacturing strategy, and the scale of the current hiring push suggests the platform is moving from development into active deployment across both sites.

The Hidden Hiring Surge: Production-Software and MES Engineers

Anduril's job board tells a story that press releases don't. While the company's public messaging centers on hardware (Roadrunner interceptors, Fury drones, Ghost drones), the hiring data points to a parallel buildout in the software layer that makes those systems manufacturable at scale. A significant slice of the listings that feed ArsenalOS aren't traditional defense manufacturing positions. They're production-software engineering roles with titles and responsibilities that look more like a scaled tech company's than a defense contractor's.

The listings are specific. A Staff Software Engineer, Production Solutions posting on Anduril's Greenhouse board describes architecting distributed systems inside Forge MES, the Manufacturing Execution System layer of ArsenalOS, covering work order management, scheduling, and material flow. A Software Engineer role calls for partnering directly with the factory floor to enable "high rate build" of Roadrunner, Fury, and other products, with an on-call rotation for Forge support on the production line. A Technical Program Manager, Digital Thread position, posted on Revolution's job board, requires someone to own the "completeness and health" of the digital thread across the organization, using digital engineering principles to drive integration.

These aren't one-off experiments. The roles span multiple locations (Costa Mesa, California; Boston; Quonset Point, Rhode Island; Santa Ana, California) and cover the full production-software stack: MES architecture, real-time production monitoring, quality intelligence data engineering, and digital thread program management. The salary ranges reflect how seriously Anduril is competing for this talent:

These numbers sit well above what traditional defense manufacturers pay for comparable manufacturing-engineering roles, and they're competitive with commercial tech companies that defense firms have historically struggled to recruit from.

The pattern is clear: Anduril is hiring production-software and MES engineers at a pace and scale that treats the software backbone as a first-class product, not a support function bolted onto a hardware line. That distinction matters, because it's exactly where the rest of the defense industry has been slow to move.

Why Software Is the Real Bottleneck in Autonomous-Weapons Manufacturing

For decades, the defense industry treated manufacturing scale as a hardware problem. Build the tooling, hire the machinists, secure the supply chain. The logic was linear and mechanical. That model worked when the product was a fighter jet with 20,000 parts and a 10-year production run. It breaks down when the product is an autonomous underwater vehicle or a collaborative combat aircraft that needs to iterate on software-defined capabilities every few months while still meeting military-grade quality and traceability standards.

The bottleneck has shifted. Hardware fabrication is still hard, but the defense sector has decades of institutional knowledge around machining composites, casting titanium, and running clean rooms. What it does not have, and what Anduril is betting billions it can build, is the software layer that ties those physical processes together: real-time production monitoring, digital thread architecture, automated quality management, and supply-chain integration that can adapt as designs change mid-production.

Acquinox Capital put the challenge bluntly in an investor analysis: "The challenge ahead is execution: proving that a software-centric company can bend metal at the scale of a traditional industrial giant." That sentence captures the gap. Anduril has raised venture capital, won a $20 billion Army enterprise contract, the largest single deal ever awarded to a venture-backed defense company, according to fed-spend.com, and built a reputation for shipping autonomous hardware faster than the primes. But none of that matters if the factory floor cannot produce at rate with the traceability the Pentagon requires.

The traceability requirement is where software becomes non-negotiable. Every component in a military system must be documented to its source material, every test result logged and retrievable, every deviation tracked through a formal quality management system. Traditional primes handle this with massive quality departments and legacy enterprise resource planning systems designed for a slower era. Anduril's bet is that ArsenalOS replaces that patchwork with a unified digital backbone, and that the engineers who build and maintain that backbone are the actual constraint on how fast the company can scale.

The job postings back this up. Zero G Talent's board shows Anduril adding roles like Data Engineer, Quality Intelligence and Metrology Engineer alongside its manufacturing-software positions. These are not shop-floor roles. They are the people building the data pipelines and measurement systems that ArsenalOS depends on. The NDT Inspector Level III posting, paying $191,000–$253,000 a year, signals how seriously Anduril treats quality data. It needs inspectors who can operate inside a software-driven workflow, not alongside one.

Compare that to Northrop Grumman's recent listings on the same board: Staff Integration Engineer, Manager Tool Engineering, Supply Chain Manufacturing Specialist. These are competent, well-paid roles, but they reflect a staffing model organized around hardware integration and supply-chain logistics. The software that connects those functions is largely assumed to exist already, bolted on from legacy vendors. Anduril is hiring as if the software is the product, because in a factory that needs to retool for a new autonomous platform in weeks rather than years, it is.

The broader defense sector is starting to recognize the shift. Analyses from WWDC Defense Con both frame the rise of companies like Anduril, Shield AI, and Helsing as a story about software-driven development meeting defense imperatives. But recognition and execution are different things. The primes have thousands of engineers; what they lack is a generation of production-software talent that thinks about manufacturing the way a DevOps team thinks about deployment: continuous, instrumented, version-controlled.

Anduril's hiring surge is the clearest signal that the company sees this gap as its primary risk. The hardware for autonomous weapons is hard. The software that proves every unit was built correctly, every material accounted for, every test passed, that is the harder problem, and the one that will determine who actually scales.

Ohio and Rhode Island: Two Factories, One Software Stack

Anduril's manufacturing footprint splits across two very different sites, but the hiring demand at both converges on the same thing: engineers who can make ArsenalOS run on the factory floor.

In Ohio, Arsenal-1 is the flagship. The 5-million-square-foot facility near Rickenbacker International Airport is designed to produce tens of thousands of autonomous systems per year, starting with the YFQ-44A Fury for the Air Force's Collaborative Combat Aircraft program. Anduril announced the Columbus, Ohio site in January 2025 and began production in March 2026, months ahead of the original July target. The company has committed nearly $1 billion to the site, with 4,000 direct jobs planned at full scale. The first 50 Ohio-based employees are already on board, and a core group of 25 technical production leads recently returned from training at Anduril's California headquarters, per Ohio Tech News.

Rhode Island plays a different role. Anduril's robotic submarine facility in Quonset Point produces the company's autonomous underwater vehicles, including the Dive-LD and Dive-XL lines. It is smaller and more specialized than Arsenal-1, but it runs on the same software backbone. ArsenalOS is designed to be product-agnostic, using common commercial machinery and processes across every autonomous vehicle Anduril builds, whether it flies or swims.

That shared stack is what makes the hiring pattern notable. A Senior Manufacturing Test Engineer position is posted for Quonset, Rhode Island, while Data Engineer and Quality Intelligence roles sit in Costa Mesa, feeding the data layer that ArsenalOS depends on. The company's open-roles page lists positions filtered by location, including Ashville, Ohio, where Arsenal-1's production ramp is accelerating.

The practical effect: Anduril needs production-software and manufacturing engineers in both states, but for different reasons. Ohio needs people who can operate and optimize a high-volume, multi-product factory running ArsenalOS at scale. Rhode Island needs engineers who can adapt the same platform to low-volume, high-complexity underwater vehicle production. Both sites need people who understand that the software layer, not the hardware, is what determines whether these factories hit their throughput targets.

Matthew Grimm, Anduril's co-founder and chief operating officer, said the production line can be reconfigured "in days and weeks, not months and years." Ohio Tech News reported that claim. It only holds if the software team is staffed to match. The hiring surge suggests Anduril knows that.

What the Job Postings Reveal About Anduril's Actual Technical Architecture

Strip away the marketing language on Anduril's careers page and the job listings themselves become a blueprint. The specific tools, workflows, and system categories named across dozens of open manufacturing and software roles sketch out what ArsenalOS actually is, and, more importantly, where it's still being built.

Start with the data layer. A Software Engineer, Data Engineering role on the Maritime Digital Production team, posted on Anduril's Greenhouse board, describes building "high-throughput data systems that connect design, manufacturing, and operations into a unified digital thread." The candidate will "design and implement scalable data architectures that move and transform massive datasets efficiently, supporting everything from geometry access to live production monitoring." That single posting confirms that ArsenalOS is, at its core, a data-integration platform, not a single monolithic application but a set of pipelines meant to unify product lifecycle data from CAD models through shop-floor execution.

The PLM and ERP integrations named in preferred qualifications reinforce this. Multiple Manufacturing Engineer listings call out experience with Teamcenter (PLM) and NetSuite (ERP) as preferred skills. A Staff Software Engineer, Production Solutions role references warehouse management systems that handle receiving, storing, moving, picking, packing, shipping, and tracing inventory. Together, these requirements point to an architecture with at least three integration surfaces: a product lifecycle management system on the design side, an ERP on the business-operations side, and a warehouse-management layer for physical material flow. ArsenalOS sits between all three, pulling data in and pushing orchestration commands out.

Quality management is another module the postings expose. Zero G Talent's board currently lists a Data Engineer, Quality Intelligence role at Anduril's Costa Mesa office. The title alone ("Quality Intelligence") signals that Anduril is building analytics on top of inspection and test data, not just collecting it. The Manufacturing Engineer listings back this up: candidates are expected to write "quality plans, inspection requirements" and "implement quality procedures" at supplier sites. The architecture has to support non-conformance tracking, statistical process control, and traceability back to individual serial numbers, the basic requirements of AS9100-compliant manufacturing, which one listing explicitly names as a preferred exposure.

Then there's the real-time production monitoring layer. The Manufacturing Engineer role at the Quonset Point AUV facility requires developing "process flows, breaking down large, complex assemblies into a logical part flow of subassemblies and sub-processes." That's shop-floor orchestration language. Someone has to build the system that tracks where each subassembly is, what work instruction applies, and whether the operator passed or failed each step. The fact that Anduril lists JIRA, a tool designed for software task tracking, as a preferred qualification alongside Teamcenter and NetSuite suggests the company is adapting agile workflow tools for hardware production tracking, a pattern more common in tech companies than in traditional defense manufacturers.

What's missing is just as telling. None of the listings reference MES platforms like Siemens Opcenter, Rockwell FactoryTalk, or Dassault DELMIA, the systems that established defense primes rely on for manufacturing execution. Anduril isn't buying an off-the-shelf MES. It's building one, or at least building the integration layer that would replace one. The "digital thread" language in the data-engineering posting is the giveaway: that term describes a still-emerging architecture in which design data, manufacturing data, and field-operations data flow through a single connected model. No legacy MES does that well. Anduril is hiring the people who will make it work inside ArsenalOS.

The supply-chain integration surface is the least mature. Manufacturing Engineer roles require candidates to "evaluate, plan, and prioritize long-term and short-term supply chain, production and test optimizations" and to "continuously evaluate make vs buy decisions for sub-assemblies." These are strategic responsibilities, not software-feature descriptions, which suggests the supply-chain intelligence layer of ArsenalOS is still largely a human-driven process that the software hasn't yet automated. The hiring push is aimed at closing that gap.

Read together, the postings describe a platform under active construction: data pipelines and digital-thread infrastructure furthest along, quality intelligence and shop-floor orchestration in progress, and supply-chain decision support still in early stages. ArsenalOS isn't a finished product Anduril is deploying. It's a hiring thesis, and the job listings are the spec.

How Anduril's Software-First Factory Approach Differs From Traditional Primes

The defense primes are still hiring. But the shape of their hiring tells a story that Anduril's leadership has been eager to exploit. LinkUp job data shows a gradual decline in open positions at legacy contractors over the past three years, even as Anduril's postings climbed. More telling is the composition: Anduril allocates roughly 52% of its job openings to computer and mathematical roles, according to LinkUp's occupational breakdown. Lockheed Martin's software hiring, by contrast, has shown no comparable surge, and no apparent urgency to compete on the same axis.

The divergence reflects a structural difference in how these companies think about manufacturing scale. Lockheed Martin and Northrop Grumman staff their production lines with the workforce they have relied on for decades: manufacturing engineers, integration specialists, hardware-focused electrical engineers, and supply chain managers. Their career pages list locations like Marietta, Georgia, and Stratford, Connecticut, alongside roles in RF engineering and flight-control actuation systems. These are real and necessary jobs. But they are not the jobs that build a digital thread through a factory floor.

Palmer Luckey made the contrast explicit after Anduril won the Air Force's CCA prototyping contract ahead of Lockheed Martin, Northrop Grumman, and Boeing. "Anduril is proving that with the right team and business model, a seven-year-old company can go toe-to-toe with players that have been around for 70-plus," he wrote on X. The "business model" he referenced is one where software is not a support function for hardware but the organizing principle of the factory itself.

The primes are not standing still. Lockheed Martin's career infrastructure and Northrop Grumman's continued hiring for advanced electronics integration show they are investing in capability. But their organizational structures, built around programs of record that span decades, make it difficult to reorient hiring toward production-software at the speed a company like Anduril can. When your manufacturing execution system was designed before cloud computing was mainstream, you cannot simply post a job for a digital-thread architect and expect the rest of the enterprise to absorb that person productively.

This is the gap that matters for engineers evaluating where to work. The primes offer stability, clearances, and programs that will outlast any single administration. Anduril offers something narrower but potentially more consequential: the chance to define how autonomous weapons manufacturing is software-defined from the ground up, rather than retrofitting that philosophy onto a factory that was already building F-35s when you were in middle school.

What This Means for Engineers Considering Defense-Tech Careers

The production-software and manufacturing-engineering talent market in defense is shifting faster than most job seekers realize. Anduril's ArsenalOS hiring is one signal, but the structural trend is broader: the aerospace and defense industry's scaling bottleneck has moved from the factory floor to the software layer that orchestrates it. Engineers who position themselves at that intersection, rather than in traditional design roles, will find the most leverage.

Deloitte's 2026 aerospace and defense outlook projects that data analysis skills will appear in nearly 14% of industry job postings by 2028, up from 9% in 2025, with data science postings growing from 3% to 5% over the same window. These are not AI-research roles. They are production-adjacent positions: engineers who can build and maintain the data pipelines, quality-intelligence systems, and digital-thread architectures that connect design to finished product. The demand is coming from primes and new entrants alike, because the problem is the same across both: legacy manufacturing execution systems were not built for the traceability and throughput rates that autonomous-weapons production demands.

The skill sets that map directly onto this gap are specific. Manufacturing-experience with MES platforms, digital-thread implementation, and real-time production monitoring. Proficiency in the data-engineering stack (Python, SQL, and cloud platforms like AWS and Azure) that defense manufacturers are increasingly adopting for production analytics. Quality-management systems and statistical process control, especially in environments that must meet AS9100 or equivalent aerospace standards. And, critically, the ability to work within cleared environments, where even production-software roles often require at least a Secret clearance and frequently TS/SCI.

Salary data from current postings reflects the premium. Northrop Grumman's staff-level integration engineers in California are posting $131,700 to $241,400. Blue Signal's 2026 hiring analysis found cleared software engineers in the defense sector are averaging $135,000 to $160,000 annually, and production-specialized engineers with domain knowledge command the upper end of that range.

The generational workforce data adds urgency. A joint 2024 report from the Aviation Technician Education Council and Oliver Wyman found the average age of a certified aircraft mechanic in the U.S. is 54, with 40% over 60. That retirement wave is hitting the production-engineering ranks too, not just the technician trades. The institutional knowledge walking out the door is exactly the kind that ArsenalOS and similar platforms are designed to capture and systematize, and someone has to build those systems.

For engineers evaluating where to place their next two to three years, the practical move is to look at the manufacturing-software and quality-intelligence roles at companies like Anduril, where production scale-up is the active constraint. The job postings are live now on Zero G Talent's board. The window is open because the problem is current, not theoretical.

Working in frontier tech? Zero G Talent tracks the openings: browse frontier tech jobs, openings at Anduril Industries and Northrop Grumman, and the people building the field.