AI Forward Deployed Engineer

Arken builds AI to take apart fraud, waste, and abuse in the US government. The product is an agentic platform that ingests public and government-restricted sources, including USA Spending, SAM.gov, agency contracting databases, and Medicare and Medicaid claims. It reasons across them and surfaces specific fraud cases with the evidence chain auditors can act on. The current focus is Medicare and Medicaid fraud, a problem on the order of $500 billion a year.

The founders have already found more than $200 million in previously unidentified fraud across three or four agencies. They are in active conversations with CMS, HHS, the Department of War, Homeland Security, and state and local governments. They have completed one funding round, with a high-powered board and strong political tailwinds. The bottleneck on multi-million-dollar contracts is headcount. Government will not send sizable money to a three-person company, so they have to staff up first.

The CEO is non-technical, a Carnegie Mellon business graduate who started a company at 14, later worked in venture and startup consulting, and has family and network in DC government contracting. The CTO started in AI research in 2017, interned at Google DeepMind in high school with a published paper, and founded an automation company that raised a $2 million pre-seed and sold about six months ago. Both founders are under 22. Seed stage, about 1 to 10 people, founded in 2025, in Washington, DC.

Arken is hiring its first cohort of AI forward deployed engineers, split about evenly across two teams. The seat is a member of technical staff spanning applied AI research and production systems, with partner-facing work inside US government environments.

The top-secret team works with non-civilian agencies, including the Department of War and Homeland Security. US citizenship is required. Engineers on that team are on site at the Pentagon two or three days a week and are sponsored for a top-secret clearance. An existing clearance is a strong plus. The civilian team works with CMS, HHS, and state and local governments. US citizenship is preferred there. An O-1 is possible for an exceptional civilian profile.

You architect and ship agentic systems from proof of concept through production. That means LLM applications, multi-agent frameworks, tool-calling agents, retrieval pipelines, and human-in-the-loop workflows. You embed with government partners, build data curation and evaluation pipelines, and keep the infrastructure for inference, experimentation, evaluation, and deployment running. You own the system from discovery through security and reliability. You have to explain the work clearly to non-technical government stakeholders.

Bar: about 1 to 4 years of production engineering. Strong Python and end-to-end delivery. Hands-on LLMs and agentic systems, not vibe-coding. A track record with data pipelines and ML infrastructure. Excellent verbal communication. The strongest fits are ex-founders or founding engineers of AI startups, forward-deployed or applied engineers at defense-adjacent companies such as Palantir, Anduril, Shield AI, or Scale's federal work, or applied engineers at agentic companies such as Sierra, Decagon, or Glean. Prior government digital-service work, an active clearance, or Medicare and Medicaid fraud experience is a plus. Internship-only, a pure researcher who has not shipped, a big-firm consultant profile, or someone who cannot relocate to DC is a miss.

In office in Washington, DC, five days a week. They want someone who can start soon. Visa sponsorship is not available. Equity is about 0.5% to 1%. Relocation support, medical, dental, and vision, and a path to top-secret clearance. Posted work intensity is 9-9-5. The process is an intro with both founders, a technical deep dive, then an in-person meeting in DC. They fly the candidate in, or a founder flies out.