We’re building a new generation of autonomous ISR aircraft capable of sensing, deciding, and acting in milliseconds, without relying on the cloud.
Your work will power the onboard AI agents that fly, detect, and coordinate missions in real time.
What you’ll do
- Build and fine-tune LLM systems that power onboard decision-making for autonomous aircraft.
- Design agentic workflows that coordinate three core AI agents onboard:
- Operator Agent – mission brain, planning and arbitration
- Pilot Agent – real-time flight and camera control
- Intel Agent – perception and verification
- Develop low-latency communication protocols between these agents to ensure reliable, deterministic behavior during flight.
- Build prompt orchestration and retrieval pipelines for situational awareness and mission control.
- Deploy optimized models on embedded hardware that handle variable telemetry in real time.
What we’re looking for
- Strong background in Python, PyTorch, and modern LLM frameworks (e.g., Hugging Face, vLLM, llama.cpp).
- Experience with fine-tuning, RAG, embeddings, or custom model adapters.
- Comfort building from research ideas to production-ready code.
- Experience working with real-time or mission-critical systems.
- Strong understanding of edge inference (quantization, optimization, GPU/CPU/NPU).
Nice to have
- Familiarity with small language models (1–3B), grammar-constrained decoding, or state-space models.
- PX4/ArduPilot/MAVLink or drone systems.
- Vision systems (OCR, lightweight detectors) or real-time telemetry handling.
- Experience with 3D simulation (Three.js, Cesium, or similar).
These agents don’t live in the cloud, they fly onboard, coordinating in real time to search, verify, and adapt.
Why USAVX
USAvionix builds autonomous systems that act faster and smarter than the threat. In this role, you’ll own one of our core technologies - shaping how autonomy is coordinated, deployed, and trusted in the field.