ML Engineer, Retrieval & Grounded Generation at DEFCON AI

Job description

ABOUT DEFCON AI RESILIENCE IN THE FACE OF DISRUPTION. DEFCON AI is an insights company that leverages artificial intelligence, mathematical optimization, data analytics, and software engineering for resilient optimization of complex systems. In today’s dynamically changing world, DEFCON AI’s technology aligns outcomes with operational goals, better decision making, and empowers customers to anticipate assess, and mitigate the impacts of disruptions. About the Role You'll join the analytics and AI engineering team behind a system that genuinely matters: an AI-assisted platform that pulls together records from dozens of external feeds, resolves them to the right person, surfaces what a human reviewer should look at first, and explains every recommendation in plain, defensible terms — running inside a secure government cloud environment. It's the kind of problem where the details you get right are the ones that count, which is exactly what makes it worth doing well. As ML Engineer, Retrieval & Grounded Generation, you'll build embeddings, vector storage, and retrieval at scale, and integrate language models so that every piece of generated text is bound to cited source records and citation failures are tested for rather than assumed away. You'll also own prompt and output-schema design; model packaging, versioning, serving, and rollback; and the telemetry hooks that make later measurement possible without manual reconstruction - real infrastructure for a real production system, not a demo. This is a fully remote role, with occasional travel to DEFCON AI HQ, customer sites, and vendor facilities as required. Key Responsibilities Implement embeddings, vector storage, and retrieval across a large, provenance-tracked evidence base Integrate language models so generated text is bound to cited source records; test for citation failures rather than assuming them away Design prompts and output schemas Own model packaging, versioning, serving, and rollback Instrument telemetry for retrieval and generation quality, recommendation/version attribution, overrides, abstentions, grounding failures, latency, throughput, and measurement events defined with Model Test Provide bounded model assistance for difficult narrative extraction where deterministic processing is insufficient, with every output tied to its source passage Supply the recorded rule context to every model-assisted step, so each output carries the exact rule versions and ordered context it received Maintain a modular in-boundary serving path, self-hosted or managed, alongside the primary managed inference service, so the platform does not depend on one provider’s availability or approval Required Qualifications 5+ years of experience, including a production or near-production retrieval-augmented (RAG) system you built yourself Ability to speak in detail to your retrieval design, which vector store you used and why, how you tested grounding, what citation failures looked like in practice, and how rollback worked Strong Python, with hands-on experience in embeddings and vector retrieval at scale Clarity on what actually shipped in past work — prototype, proposal, or deployed code — since that distinction matters more here than the title on a resume US Citizenship Required Active US Secret clearance required to start. Preferred Qualifications Experience deploying models into restricted or air-gapped environments Self-hosted or open-weight model operation Fine-tuning, adapters, or custom embeddings Federal DevSecOps, RMF, ATO, or DoW cloud environment experience Active Top Secret clearance What Success Looks Like Generated explanations that assert no more than the sources support, with the citation path intact and citation failures tested rather than assumed away A retrieval system that performs at scale on a large, provenance-tracked evidence base Model rollback that works when it's needed, with telemetry complete enough that measurement does not require manual reconstruction What We Offer A fully remote, results-based environment Competitive salary, bonus, and equity package 100% employer paid, comprehensive health insurance including medical, dental, and vision for you and your family Unlimited PTO, with your manager's approval Flexible work environment where you manage your work day 14 weeks of fully-paid parental leave Salary Range: $165,000–$200,000. This represents the typical salary range for this position based on experience, skills, and other factors. We’re an Equal Opportunity Employer: You’ll receive consideration for employment without regard to race, sex, color, religion, sexual orientation, gender identity, national origin, protected veteran status, or on the basis of disability. Applicant Data Disclosure By submitting an application, you acknowledge that Defcon AI uses third-party service providers to facilitate its recruitment and hiring processes. These providers include applicant tracking systems, candidate verification platforms, and fraud detection tools (collectively, "Hiring Platforms"). Your application materials, including your résumé, cover letter, work samples, responses to application questions, and any other information you submit, may be transmitted to and processed by these Hiring Platforms for the following purposes: Managing and administering your application throughout the hiring process; Verifying the accuracy and authenticity of application materials, including by cross-referencing information you provide against publicly available sources and proprietary databases; Identifying indicators of potentially fraudulent, fabricated, or materially misleading application content, including but not limited to discrepancies between submitted materials and publicly available professional profiles, geographic anomalies, and fabricated work histories. Applications that are flagged through this process as containing indicators of fraud or material misrepresentation may be declined from further consideration. If you have questions about the status of your application or the evaluation process, please contact recruiting@defconai.com . Defcon AI requires its Hiring Platform providers to process your information solely for the purposes described above and in accordance with applicable law. Your information will be retained only for as long as necessary to fulfill these purposes and any applicable legal obligations, after which it will be deleted in accordance with Defcon AI's data retention policies. For more information about how your data is used, please refer to our Privacy Policy and Applicant Privacy Notice .

Requirements

Must-haves

Implement embeddings, vector storage, and retrieval across a large, provenance-tracked evidence base
Integrate language models so generated text is bound to cited source records; test for citation failures rather than assuming them away
Design prompts and output schemas
Own model packaging, versioning, serving, and rollback
Instrument telemetry for retrieval and generation quality, recommendation/version attribution, overrides, abstentions, grounding failures, latency, throughput, and measurement events defined with Model Test
Provide bounded model assistance for difficult narrative extraction where deterministic processing is insufficient, with every output tied to its source passage
Supply the recorded rule context to every model-assisted step, so each output carries the exact rule versions and ordered context it received
Maintain a modular in-boundary serving path, self-hosted or managed, alongside the primary managed inference service, so the platform does not depend on one provider’s availability or approval
5+ years of experience, including a production or near-production retrieval-augmented (RAG) system you built yourself
Ability to speak in detail to your retrieval design, which vector store you used and why, how you tested grounding, what citation failures looked like in practice, and how rollback worked
Strong Python, with hands-on experience in embeddings and vector retrieval at scale
Clarity on what actually shipped in past work — prototype, proposal, or deployed code — since that distinction matters more here than the title on a resume
US Citizenship Required
Active US Secret clearance required to start.

Nice-to-haves

Experience deploying models into restricted or air-gapped environments
Self-hosted or open-weight model operation
Fine-tuning, adapters, or custom embeddings
Federal DevSecOps, RMF, ATO, or DoW cloud environment experience
Active Top Secret clearance
What Success Looks Like
Generated explanations that assert no more than the sources support, with the citation path intact and citation failures tested rather than assumed away
A retrieval system that performs at scale on a large, provenance-tracked evidence base
Model rollback that works when it's needed, with telemetry complete enough that measurement does not require manual reconstruction

Application prep

Use the role’s language and application details while tailoring your materials.

Keywords to mirror

aimlpython

Platform: greenhouse
Effort: quick

Benefits

Benefits detected in the listing.

Health Insurance

Dental

Vision

Pto

Parental Leave