Forward Deployed Engineer, U.S. Government
Job Type: Full-time
Location: Based in DC
The Role
We're hiring an MTS to work on applied AI systems across research and production. This role spans model experimentation, infrastructure, and deployment in high-assurance, mission-critical environments, including forward-deployed work with strategic partners to ensure mission success.
What You'll Work On
- Build and deploy agentic AI systems that support high-impact U.S. Government missions from prototype to production.
- Design and integrate LLM applications, including multi-agent systems, tool-using agents, RAG workflows, and human-in-the-loop autonomy.
- Work directly with government partners in forward-deployed environments to translate mission needs into technical systems.
- Build and maintain large-scale data curation and evaluation pipelines for model training and improvement.
- Develop infrastructure for model inference, experimentation, evaluation, and deployment across frontier AI platforms.
- Build AI systems that turn mission data, models, and tools into real operational autonomy.
- Own systems across the full lifecycle, including discovery, architecture, implementation, deployment, reliability, security, and iteration.
What We're Looking For
- Able to operate independently in ambiguous, partner-facing settings with technical and operational ownership.
- Strong Python engineer with end-to-end ownership experience.
- Experience with LLMs and creating agentic systems.
- Built or maintained data pipelines and ML infrastructure.
- Familiar with RL workflows or simulation environments.
- Comfortable in high-security, high-reliability environments.
Preferred Qualifications
- Active U.S. security clearance
- Experience supporting U.S. Government, defense, intelligence, or national security missions.
- Prior experience at a startup, defense organization, Department of War, or government-focused technology company.
- Background operating in fast-moving startup environments, mission-critical government settings, or both.
