Anduril
Software Engineer, Agent Platform
$220K–$292K
Washington, District of Columbia, United States +1 · Full-time · Posted Jun 26, 2026
- Location
- Washington, District of Columbia, United States
Costa Mesa, California, United States
About the job
Build and evolve Anduril’s internal LLM agent platform so teams can develop, evaluate, and deploy reliable agents in mission-critical settings.
What you'll do
- Own the agent framework’s architecture, runtime, integrations, developer experience, and reliability.
- Enable partner teams to build agents through platform capabilities, guidance, and architectural reviews.
- Integrate model post-training workflows and develop evaluation tooling for agent performance and deployment readiness.
What they ask for
- Strong backend engineering experience building production-quality platforms, frameworks, APIs, or infrastructure used by engineers.
- Deep expertise in LLM agent framework design and orchestration tradeoffs.
- Experience designing agent evaluation paradigms, including metrics, testing, and human review.
- Familiarity with model post-training workflows, plus sound judgment about reliability, observability, debugging, and safety.
Nice to have
- Experience with agent frameworks like Langchain Deepagents, Claude SDK, etc.
- Experience with evaluation platforms, simulation environments, benchmark suites, agent test harnesses, Kubernetes, Docker, distributed systems, workflow orchestration, or ML infrastructure.
Pay
$220,000 — $292,000 USD (base salary)
Tools
LLM, Langchain Deepagents, Claude SDK, Kubernetes, Docker
AI focus
Develop and evaluate LLM agent frameworks; integrate supervised fine-tuning, preference tuning, reinforcement learning, and environment-based agent training; support agent evaluation and reliability in high-stakes settings.
Model-written summary · the original posting is authoritative