Careers

AI Software Engineer Jobs with a Security Clearance

Build the government-owned reasoning infrastructure that turns fragmented national-security data into decisions at machine speed.

AI software engineer jobs in defense build the software that ingests, fuses, and serves mission data to AI and to the people who rely on it, usually for employers whose work requires a U.S. security clearance or the ability to obtain one.

By the numbers. BLS projects employment of software developers, quality assurance analysts, and testers to grow 10 percent from 2025 to 2035, with about 106,100 openings a year (BLS). In defense, speed is the gap: as of June 2021, only 6 of 36 weapon programs that reported using Agile also reported delivering software to users in less than 3 months (GAO-21-105298).

This page is for software engineers considering defense and national security work. It covers what the job is, why the department is pushing for faster software, how the work differs from commercial engineering, what it pays nationally and in Kansas City, and what it looks like at Torch.AI.

What does an AI software engineer do in defense?

The Department of War (DoW, formerly the Department of Defense) defines a Software Developer work role (code 621) in its Cyber Workforce Framework that "executes software planning, requirements, risk management, design, development, architecture, modeling, estimation, configuration management, quality, security, and tests" across all lifecycle phases. A related Software/Cloud Architect role (628) covers "application design, cloud computing strategy and adoption, and integration of software applications into a functioning system" (DCWF Tool v5.3, Sept. 2026).

At Torch.AI, software engineers do not write CRUD over a database. They build the systems that ingest, understand, and fuse the messiest data in the world, and make it usable for the people defending the country:

  • High-throughput data processing. In-memory pipelines that extract, tag, and fuse structured and unstructured data in real time, at a scale and speed where the usual tools fall over.
  • Systems that run where the network does not. Software that keeps working in denied, degraded, intermittent, and limited (DDIL) conditions at the tactical edge, disconnected from any cloud.
  • Government-owned software. Code and logic the customer owns, inspects, and evolves themselves, built for transparency and auditability rather than lock-in.
  • Integration without rip-and-replace. Connectors and bridges that meet mission systems where they already live.
How it works

From messy mission data to software that ships

DOCUMENTS, IMAGERY, AND FEEDSANSWERS IN EXISTING MISSION SYSTEMSINGESTBring in documents,imagery, and feedsEXTRACTPull text andentities from everyformatFUSELink structured andunstructured datain real timeSERVEDeliver intosystems thecustomer alreadyrunsRUNKeep working whenthe network dropsCode the customer owns, inspects, and evolves themselves.

Defense software engineering is data engineering under hard constraints: high-throughput fusion of messy data, delivered into existing mission systems, and able to run at the disconnected edge.

The department wants this delivered faster: programs on the software acquisition pathway must show operational viability within one year of first funding (10 U.S.C. 3603).

Loops · respects reduced-motion · illustrative

Why does defense need faster software delivery?

Because the department's own reviews say software is too slow to reach users. GAO found that "only six of 36 weapon programs that reported using Agile also reported delivering software to users in less than 3 months," and that "over half of the weapon systems and nearly all major business IT programs GAO reviewed reported staffing challenges" related to software development (GAO-21-105298).

The policy response has been to make speed the default:

  • The law sets a one-year clock. Programs on the software acquisition pathway must demonstrate "the viability and effectiveness of such capabilities for operational use not later than one year after the date on which funds are first obligated" (10 U.S.C. 3603).
  • The software pathway is now preferred. In March 2025 the Secretary issued a memorandum directing personnel to "adopt the Software Acquisition Pathway as the preferred pathway for all software development components of business and weapon system programs" (GAO-23-105867, recommendation status).
  • Software talent is named in law. Section 909 of the FY2022 National Defense Authorization Act "defines digital talent as including positions and capabilities in, or related to, AI, data science, autonomy, and software development" (GAO-24-105645, p. 12).

For an engineer, that means the work is wanted, funded, and measured on how fast it reaches the people who use it.

How is defense software engineering different from commercial work?

The languages and tools are familiar. The constraints are what change. The comparison below is Torch.AI's view, anchored to the public standards that apply.

Dimension Typical commercial software Defense and national security software
Where it runs Public cloud, always connected Classified enclaves and disconnected or degraded networks at the edge
Data Mostly first-party, one classification Fragmented across systems and classification levels
Secure development Varies by team Expected practice: NIST's Secure Software Development Framework, "a core set of high-level secure software development practices" for every life cycle
AI models Often treated like any dependency NIST SP 800-218A adds secure practices "specific to AI model development throughout the software development life cycle"
Ownership Vendor-owned, often closed Increasingly government-owned systems the customer can inspect and evolve
Delivery path Continuous release The software acquisition pathway, with a one-year operational viability clock

Sources: NIST SP 800-218; NIST SP 800-218A; 10 U.S.C. 3603.

Ship software to the tactical edge.

High-throughput pipelines and DDIL-ready systems the customer owns. Find the engineering role for you.

No CRUD.
Messy, real data.
Code the customer owns.

See engineering roles >

High-throughput pipelines and DDIL-ready systems the customer owns. Find the engineering role for you.

No CRUD.
Messy, real data.
Code the customer owns.

What do AI software engineer jobs pay?

BLS has no separate code for AI software engineers; the work sits inside the occupations below. The table compares national medians with the Kansas City metro, where Torch.AI is headquartered.

Occupation (SOC) U.S. jobs U.S. median wage KC metro jobs KC median wage
Software developers (15-1252) 1,687,890 $135,980 12,160 $124,990
Computer occupations, all other (15-1299) 435,370 $116,580 3,900 $105,470
Software QA analysts and testers (15-1253) 186,740 $104,300 1,430 $94,970

Source: BLS Occupational Employment and Wage Statistics, May 2025, national and Kansas City MO-KS MSA (28140). These are occupation-wide medians, not Torch.AI salaries.

  • Software is a deep local market. The Kansas City metro employs more than 12,000 software developers, slightly more concentrated than the nation (location quotient 1.02), and "computer occupations, all other" is 27 percent more concentrated there (location quotient 1.27, BLS OEWS).
  • AI is named as a demand driver. BLS projects strong demand for software developers, QA analysts, and testers due in part to "continued expansion of software development for artificial intelligence (AI)."
  • For more on the local market, see Kansas City AI jobs.

What skills do software engineers need for defense AI work?

  • Data pipelines at scale. Extracting and fusing structured and unstructured data in real time is the core of the work.
  • Distributed and edge systems. Designing for intermittent networks and limited hardware, not just elastic cloud.
  • Secure development by default. Practices like those in NIST's SSDF, and for AI components, the model-specific practices in SP 800-218A.
  • Integration. Building connectors into systems the customer already runs, instead of replacing them.
  • Clear ownership. Writing code a government customer can read, maintain, and audit after handoff.

Do AI software engineer jobs require a security clearance?

Many do. Under 32 CFR 117.10, the government determines eligibility, a contractor may request it only when classified access "is essential" to a contract, and requests "will not be used to establish a cache of cleared employees." A contractor can start the process after a written offer is accepted in writing, before your first day. At Torch.AI, many roles require an active U.S. security clearance; others are open to clearable U.S. persons, with sponsorship for the right engineer. The step-by-step path is in Kansas City AI jobs.

What is it like to build software at Torch.AI?

Torch.AI builds government-owned reasoning infrastructure that turns fragmented national-security data into decisions at machine speed. The work is hard on purpose, and it ships to real missions. Torch.AI is headquartered in Leawood, Kansas, with engineering roles there and in the Washington, DC region (see current openings).

Is this you? You will likely do well here if you:

  • Want real ownership and real impact, and want to see your work reach a mission, not a backlog.
  • Like hard problems on messy, real-world data more than clean problems on toy data.
  • Care that the mission matters, and want to build systems the customer actually owns.
  • Hold or can hold a U.S. security clearance.

This may not be for you if you want a pure-research sandbox with no delivery pressure, you are optimizing only for compensation, or you would rather stay far from the mission the software serves.

Related paths: AI and Machine Learning Engineer, AI Data Engineer, and DevOps and DevSecOps Engineer.

How do you apply for AI software engineer jobs at Torch.AI?

Current openings, locations, and clearance requirements are listed at torch.ai/careers. Each listing states whether the role requires an active clearance or is open to clearable candidates.

Build reasoning infrastructure.

Hard problems, real missions, and software that runs where the network does not. Start here.

Real ownership.
Real impact.
Real missions.

View open roles >

Hard problems, real missions, and software that runs where the network does not. Start here.

Real ownership.
Real impact.
Real missions.

Sources

See open roles