A frontier AI company is building systems that can act in the physical world, experimenting, engineering, and executing multi-step processes with real-world constraints. Backed by major research funding and operating at the edge of physical-AI innovation, they’re creating capabilities that don’t exist anywhere else. Join to work from first principles, own high-impact systems end-to-end, and help define how agentic AI will operate complex workflows in the real world.
Why This Role Matters
- Build agent systems that plan, execute, and recover across intricate engineering workflows
- Shape foundational behaviour patterns for next-gen LLM tool-use
- Join early enough to influence architecture, culture, and performance standards
- Work on problems that sit far beyond typical “LLM app” engineering
What You’ll Do
- Develop planners, state machines, and tool-calling flows using frameworks like LangGraph
- Create schemas, action definitions, and cross-tool interfaces for reliable, traceable execution
- Build error-handling, timeouts, retries, rollbacks, and replay mechanisms
- Partner with ML, infra, and systems teams to integrate agents into real engineering toolchains
What You Bring
- Strong experience with agent systems, structured tool calling, or orchestration frameworks
- Deep intuition for schemas, deterministic execution, and multi-step workflow design
- Ability to model failure modes, edge cases, and safe interactions in complex systems
- Comfort working across AI, systems engineering, and specialised domain tools in a high-precision environment
