At NEURA Robotics, cognitive robots don't just execute commands. They perceive, reason, plan over long horizons, and act autonomously in the physical world. Our robotic foundation models handle the physical execution. What sits above them is the agent: the part that works out what a task actually requires, breaks it into steps the robot can carry out, and notices when reality has stopped matching the plan. We're looking for an engineer to help build those agents for our humanoid platform.

Your Mission & Challenges

• Embodied agent harness: Design and build the agent that runs on our humanoid platform — the loop from sensor input to reasoning to action, its interfaces to the robot's cameras, depth, and proprioception, and the state management that keeps a task coherent over minutes rather than seconds.

• Task decomposition & long-horizon planning: Turn high-level human intent into ordered, executable steps, each with the preconditions, spatial references, and success criteria it needs, and sequence them so the robot can work through a task it has never seen in that exact form before.

• Visual reasoning & world state: Use VLMs to interpret what the robot is looking at, ground language in the scene, and maintain a working representation of objects, affordances, and task history for the planner to reason over.

• Closed-loop execution & recovery: Verify each step against observation instead of assuming success, detect divergence from the plan, and replan, retry, or escalate to a human when the risk is too high.

• Evaluation on the robot: Measure task success in simulation and on real hardware, build failure taxonomies from real runs, and use them to decide whether a fix belongs in the planner, the interface, or the step being issued.

• Cross-functional collaboration: Work closely with our foundation model, perception, manipulation, and hardware teams to take the reasoning layer above the models from concept to reliable behavior on a deployed robot.

What We Can Look Forward To

• A strong Master's or PhD in Computer Science, Robotics, or a related field — or equivalent depth shown through shipped work.

• Demonstrated experience building agent systems on top of LLMs and VLMs that decompose tasks, plan, use tools, and maintain state over long horizons, rather than single-turn prompt work.

• Several years of hands-on work on physical robots, not only in simulation: bringing your own code onto a real platform, running it there, and debugging a system that behaved differently once it had a body.

• Hands-on experience with VLMs as a working tool: grounding language in images, spatial and physical reasoning, structured output you can act on, and a realistic sense of where they fail.

• Familiarity with hierarchical autonomy, where a reasoning model plans and learned or classical policies execute: subgoal interfaces, progress verification, replanning.

• Experience with robotics middleware and onboard deployment: ROS 2, inference on embedded GPUs, and working inside a real latency budget.

• Solid software engineering: clean code, testing, and services other teams can build against. Cloud, containerization, and CI/CD experience is a plus.

• Excellent English. German is a plus.