Applied LLM Systems Engineer

Team Overview

Anduril’s Technical Publications organization supports complex hardware, software, autonomy, and defense systems across multiple divisions, business lines, and programs. The Product Manager, Technical Writing Systems, owns the authoring and publishing platform. This role builds the AI systems that extend that platform so the documentation pipeline handles more work at lower cost and lower latency.

Role Specifics

As the Applied LLM Systems Engineer, your mission is to design, build, and operate production AI systems that improve how technical documentation is created, transformed, validated, maintained, and delivered. You will build and maintain AI-assisted authoring, automated publishing, and multi-agent coordination tooling for the Technical Publications pipeline. This is not a prompt-writing role. It is a systems-engineering role for someone who has already delivered production AI systems, understands evaluation and rollback, and knows how to make probabilistic systems useful in enterprise environments with structured content, strict review requirements, security boundaries, and real operational consequences. The right candidate can expand system capability while managing token usage, context efficiency, and cost at scale.

What You’ll Do

• Build production-grade LLM applications for documentation and knowledge-work workflows.
• Design orchestration for multi-step workflows that coordinate models, tools, and deterministic services safely.
• Optimize context management, token usage, caching, and workflow design for cost, latency, and task success.
• Build retrieval, structured-output, and tool-integrated systems that safely interact with enterprise content, source control, issue tracking, and documentation systems.
• Design evaluation frameworks and regression testing for prompts, models, tools, retrieval pipelines, and end-to-end workflows.
• Establish observability, auditability, rollback, and safe failure modes for AI-assisted workflows.
• Work directly with technical writers, illustrators, engineers, and documentation leadership to identify the highest-value places to apply AI and where human judgment should remain primary.
• Preserve provenance between source material, generated output, validation results, and human decisions so outputs remain reviewable, correctable, and attributable.

Required Qualifications

• 5+ years of professional software engineering experience, including building and operating production services.
• Direct experience delivering at least one production AI system with measurable adoption by its intended users.
• Experience optimizing cost, latency, and context usage in production LLM or workflow systems.
• Experience designing evaluation, observability, rollback, and failure-handling for nondeterministic systems.
• Strong proficiency in Python and modern API, data, and service-development practices, including software architecture, security boundaries, identity and access management, and data protection.
• Sound judgment about where probabilistic AI is appropriate and where deterministic software or human review is required.
• Must be eligible for Secret and higher security clearance (US).
• In this role, you will be subject to pre-employment and randomized substance screening, as required by the Company.

Preferred Qualifications

• Experience designing multi-step orchestration or coordination systems across models, tools, and deterministic services.
• Experience with retrieval-augmented generation, structured outputs, tool calling, or other production LLM workflow patterns.
• Experience integrating AI capabilities into desktop or enterprise applications with limited native APIs.
• Experience in defense, aerospace, manufacturing, robotics, autonomy, aviation, or another regulated environment.
• Familiarity with technical documentation, structured authoring, content management, or comparable documentation-heavy workflows.
• Familiarity with S1000D, DITA, MIL-STD-40051, or comparable documentation frameworks.
• Experience with containerized services, cloud infrastructure, CI/CD, queues, event-driven architectures, or production monitoring.