About Binabik AI
Binabik AI is a deep-tech robotics startup with roots at ZHAW, building the first agentic operating system for humanoid and industrial robots: a modular "brain" that lets robots learn quickly from limited data and perform long-horizon tasks in dynamic, unstructured environments.
Our stack is ROS 2 native and orchestrates specialized models, reusable skills, and persistent memory through a multi-agent architecture. This is what lets our robots adapt to new tasks with far less data than end-to-end approaches.
We are based at Technopark Winterthur and backed by a strong network of grants, strategic hardware partners across the leading humanoid manufacturers.
About the Role
As a Robotics Research Engineer, you develop the core software of Binabik AI: the agentic physical AI operating system that powers humanoid robots. You will work across manipulation, ROS-based systems, and learning components, and help turn research results into a reliable product that runs on real hardware.
Responsibilities
Develop and maintain the core software behind Binabik's agentic operating system.
Work on robotic manipulation and the integration of our software with humanoid platforms.
Contribute to data collection, teleoperation, vision-language-action models, and agentic AI components.
Move research results into robust, deployable software in close collaboration with the team.
Test and validate your work on physical robots.
Requirements
An MSc in robotics, computer science, AI, or a related field (a PhD is a plus).
Strong programming skills in Python; comfortable in a (distributed) Linux environment.
Hands-on experience with ROS or ROS 2, including building and debugging real robotic systems.
Demonstrated ability to debug across the full stack — from a misbehaving sensor or actuator up to a misbehaving policy or planner.
Experience with robotic manipulation.
Fluent English, which is our working language.
The right to work in Switzerland.
Nice to Have
Experience with humanoid or mobile-manipulation platforms (Agibot, Unitree, Galaxea, or similar) which is a strong plus.
Experience with teleoperation systems, imitation learning, or data-collection pipelines for robot learning.
Familiarity with manipulation stacks, motion planning, or modern perception models (VLAs, VLMs in a robotics context).
A strong publication record at leading venues. In robotics, this includes RSS, CoRL, ICRA, IROS, and IEEE-RAS Humanoids. In AI and machine learning, this includes NeurIPS, ICML, ICLR, and CVPR.
German.
What We Offer
A core engineering role at an early-stage company building the brain of humanoid robots.
Real ownership of central parts of our software and direct technical impact.
Direct work with state-of-the-art humanoid robots.
Close collaboration with a founding team that combines robotics and AI research.
