SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Sereact is a Series B embodied-AI company deploying Vision Language Action (VLA) models in live warehouse robotic cells across the EU and US. This Research Engineer role is based in the Zurich research hub and owns the data generation layer of Sereact's VLA stack.
You will build and operate teleoperation systems for dexterous manipulation, including retargeting algorithms that map human hand and arm motion onto robot end-effectors. You'll design and maintain scalable, UMI-style data collection hardware and software that other engineers at Sereact rely on. Your responsibilities include benchmarking control and retargeting approaches using rigorous, human-factors-informed methodology; instrumenting and supporting physical robots deployed in the Zurich hub, including day-to-day work on the control stack; and working closely with the model-training team to define data requirements and close the loop on data quality. You'll also build tooling or infrastructure outside your immediate brief when it unblocks the wider team.
This is an individual contributor role in a small, still-forming team. You'll have direct collaboration with founders and world-leading AI experts, access to large-scale compute clusters and proprietary robotic datasets, and the opportunity to shape the future of embodied AI.
REQUIREMENTS:
- MSc (or equivalent, in progress) in robotics, mechanical engineering, or closely related field, with thesis or capstone project on manipulation, teleoperation, or retargeting
- Hands-on experience building a data collection device or system for robot learning (UMI or comparable system), not just operating one someone else built
- Working knowledge of retargeting algorithms for mapping human input onto robot hands or end-effectors
- 1 to 2 years of combined research, internship, or industry experience in robotics, mechatronics, or related engineering discipline
- Software and systems engineering skills strong enough to build and maintain infrastructure other engineers rely on, not only research-grade prototypes
- Comfortable working as an individual contributor in a small, still-forming team
NICE TO HAVE:
- Conference submission on manipulation, teleoperation, or retargeting where candidate is an author (MSc thesis submitted to venue like CoRL counts)
- Teaching or TA experience in robotics course
- Exposure to large-model training
- Concrete example of building something outside formal role brief because team needed it
