Agentic AI Engineer (Automation)

• Air Dominance & Strike designs, builds, and flies autonomous air vehicles — from collaborative combat aircraft to expendable cruise missiles to counter-UAS interceptors
• Our vehicles move from whiteboard to first flight on timelines that traditional primes consider impossible, which means our design cycles live or die on how fast we can close the iteration loop and begin testing
• The Anduril AI-Engineering team exists to collapse that loop
• We are engineers first
• We work from engineering first principles and unlock capability through machine learning
• We are building to scale across design, analysis, test, and program execution, with agent pipelines and tooling that carry across programs
• We are looking for an Agentic AI Engineer to automate engineering workflows, build new capability through agents, and improve how engineers access and act on their data
• Our R&D work runs through tools of varying fidelity, from empirical methods and low-order models to high-fidelity solvers
• You will chain those tools into automated flows that carry a design through analysis to build and then back into the next iteration
• You will own agent pipelines end to end: orchestration logic, tool and data integrations, evaluation harnesses, and guardrails
• Much of the work is getting agents to drive legacy engineering software which were built for human operators rather than programmatic control
• You will also use agents to build new tools as programs evolve and requirements change, standing up capability on program timelines rather than software release cycles
• Defense experience is not required
• We are looking for engineers who came to machine learning through the problems they were already trying to solve
• Build and deploy multi-agent pipelines that compress the engineering design loop, including automated case setup, batch submission, and post-processing of solver runs across CFD, FEA, thermal, and electromagnetics
• Design, implement, and evolve how agents interact with classical engineering software: tool calling, prompt engineering, task decomposition, state management, retries, and human-in-the-loop checkpoints,
• Automate data extraction and aggregation across solvers, test benches, and program systems, and build the dashboards that give engineers and leadership rapid access to their own results
• Build bespoke tooling through agents for engineering sub-disciplines, including aerodynamics, thermal, GNC, structures, and avionics, that brings new capability in-house
• Choose which model to run at each stage of agent work, from planning through tool development to execution, and tune routing, context management, and caching so agents run efficiently against token cost and latency budgets
• Build the evaluation harnesses, guardrails, and sandboxed execution needed to defend agent behavior before it is deployed on a program
• Mentor engineers who are not ML specialists, and work with them to identify and scope the projects where agents deliver the highest impact

Benefits

• Full Family Health Coverage
• 16 Weeks Paid Parental Leave for All Caregivers
• Family Planning & Support
• Incentivized Time Off
• Mental Health Resources
• Financial Planning
• Unlimited Provisions
• Professional Development- Strong programming skills in Python and MATLAB, and working knowledge of at least one additional core language such as Java, Go, or C++
• Eligible to obtain and maintain a U.S. Secret security clearance
• Experience with data extraction, aggregation, and sanitization techniques, and with building production data pipelines from heterogeneous engineering and test sources
• MS or PhD in aerospace, mechanical, or electrical engineering, computer science, data science, or machine learning
• 0-3 years of professional experience, with demonstrated hands-on work implementing agentic AI systems
• Proficiency developing on Linux, with containerized deployment via Docker and Kubernetes
• Experience building agentic systems, including multi-agent orchestration, tool calling, prompt engineering, integration into classical software systems and retrieval-augmented generation
• Experience with Model Context Protocols, including building and maintaining MCP servers
• Experience deploying LLM applications in accredited or otherwise restricted cloud environments
• Experience with cost and latency optimization at scale, including caching, batching, and prompt and context efficiency
• Experience with structured output, function calling, and prompt optimization at production scale
• Familiarity with engineering toolchains such as PLM systems, requirements management tools, solvers, and test data systems
• Experience building knowledge graphs or semantic layers over engineering data