Robotics Engineer - Learning for Manipulation
Bota Systems · Zürich, CH · Posted Nov 7, 2026
Transparency
StaleOnline for 13 days and never reposted. The first wave of applicants is in, so a reply is less likely. Apply if it's a strong fit.
- First seen
- Nov 7, 2026
- Offered for
- 13 days
Overview
- Category
- AI & machine learning
- Employment type
- Permanent
- Workplace type
- On-site
- Country
- Switzerland
- Seniority
- Junior
- Schedule
- Full time
- Salary
- Not listed
Key skills
- Learning-based Manipulation Policy Training
- Train, fine‑tune and deploy imitation, diffusion and VLA policies using vision and force/torque inputs.
- Multi‑modal Data Collection Pipelines
- Build and operate kinesthetic teaching, teleoperation and synchronized recording pipelines for dataset creation.
- Force‑aware Simulation Integration
- Integrate physics‑consistent simulation that respects force feedback for training and evaluation.
- ROS 2 and Real Robot Integration
- Develop software that connects ROS 2 nodes, sensors and robot hardware reliably and at scale.
- Python and C++ Development
- Write and maintain high‑performance code in Python and C++ for learning stacks and robot control.
Key phrases
- learning-based manipulation policies
- imitation learning
- diffusion policies
- VLAs
- vision and force/torque
- data-collection pipelines
- kinesthetic teaching
- teleoperation
- synchronized multi-modal recording
- dataset curation
- force-aware, physics-consistent simulation
- ROS 2
- real robot hardware
- LeRobot
- MuJoCo
From the posting
At Bota Systems, we build force/torque sensors and the tools that let robots learn from touch. We’re now building a platform for training robot AI that combines vision and force sensing, and we’re looking for an engineer to help robots learn contact-rich manipulation from real-world data.
Position: Robotics Engineer - Learning for Manipulation
Location: Zurich, Switzerland
Start Date: As soon as possible
What you’ll do:
- Train, fine-tune, and deploy learning-based manipulation policies (imitation learning, diffusion policies, VLAs) that use both vision and force/torque as inputs
- Build and run data-collection pipelines on real hardware, including kinesthetic teaching, teleoperation, synchronized multi-modal recording, and dataset curation
- Integrate force-aware, physics-consistent simulation for training and evaluation
- Design the software architecture that connects sensors, robots, and learning stack, with a focus on reliability and scalability
- Work directly with our hardware and embedded engineers to bring models from experiment to working robot
What you bring:
- A degree in Robotics, Computer Science, Electrical Engineering, or a related field, ideally 1–3 years of relevant experience (industry, research, or a substantial thesis)
- Hands-on experience in training and fine-tuning ML models, ideally for robotics
- Strong Python skills and working knowledge of C++
- Experience with ROS 2 and with real robot hardware
- Clear communication and enjoyment of working across disciplines
Nice to have:
- Experience with robot learning frameworks (e.g. LeRobot), policy architectures such as ACT, diffusion policy, or VLAs, or multi-modal/tactile learning
- Simulation experience (MuJoCo, Isaac Sim/Lab, or similar)
- Experience with force control, impedance control, or contact-rich tasks
What we offer:
- Real ownership of a core part of our product, with direct impact on what we ship
- Daily access to state-of-the-art sensors, robots, and teleoperation setups
- A small, highly technical team in Zurich
Benefits:
- Opportunity to work on exciting projects at the forefront of robotics.
- Mentorship from experienced professionals in the field.
- Hands-on experience with robotic systems and technologies.
Apply with your CV and a short note on a manipulation or robotics project you’re proud of.
More jobs at Bota Systems
See all 9 jobs at Bota Systems
