Job Description

We are looking for a Master’s student to conduct their thesis within our prototyping and applied systems team, contributing to research at the intersection of robot learning, multimodal data, and brain–computer interfaces.

Robot demonstrations capture what a person does, but not necessarily when they recognize an error, reconsider an action, or initiate a correction. We investigate whether EEG-derived annotations can identify demonstration segments that are particularly useful for training autonomous robots.

Working closely with researchers and engineers, you will tackle a focused validation question:

can neural-informed selection or weighting of demonstration data improve robot policy performance or reduce the amount of training data required?

The project combines egocentric or teleoperation recordings, framewise behavioral annotations, and densely sampled mental-state classifier outputs. Your work will focus on downstream robot-learning experiments, with support from our neuroengineering team for interpreting the neural signals.

The role is intended to align with a Master’s thesis, subject to topic fit and university supervision.

Start: By agreement, as soon as possible

Duration: 4–6 months

Availability: 10-20 hours per week

What you will do

• Define a tractable research question and evaluation protocol with the team
• Prepare aligned demonstration data and annotations for robot-policy training
• Implement methods for selecting or weighting demonstrations using neural and behavioral information
• Compare against random selection and non-neural baselines, potentially including policy-influence methods
• Train and evaluate robot policies in simulation and/or on available hardware
• Assess task success, generalization, and data efficiency using controlled experiments and statistical analysis
• Document reproducible experiments and contribute to your thesis and potential research publications

The thesis will address one well-defined validation step rather than require building the complete recording, neural-analysis, and robot-learning pipeline.

What we offer

• Hands-on research connecting human neural signals with robot learning
• Close collaboration with BCI researchers and applied-systems engineers
• The opportunity to develop a rigorous thesis with practical relevance
• A pragmatic, low-hierarchy environment with room to take ownership
• Flexible scheduling around exams and coursework
• Support for scientific writing and potential dissemination of the results

Requirements

What you bring:

• Current enrollment in a Master’s program in Robotics, Computer Science, Machine Learning, AI, or a related field
• Strong interest in robot learning, imitation learning, or learning from demonstrations
• Solid Python programming skills and experience with a machine-learning framework such as PyTorch
• Familiarity with training and evaluating machine-learning models
• An understanding of experimental design, data leakage, and meaningful baseline comparisons
• Ability to translate an open research question into a manageable implementation and evaluation plan
• Comfortable working independently while collaborating across disciplines

Nice to have:

• Experience with imitation learning, reinforcement learning, or vision-language-action models
• Familiarity with robot-learning tools such as LeRobot, RoboMimic, or simulation environments
• Experience with multimodal time-series data, dataset curation, or data valuation
• Experience running reproducible experiments across multiple training seeds
• Exposure to EEG or brain–computer interfaces; prior neuroscience experience is not required
• Prior involvement in research projects, open-source contributions, or scientific writing