Bram Grooten

Postdoc in Deep Learning at the TU Eindhoven

CV.pdf

I am a postdoc in the the AMOR/e lab at the TU Eindhoven, working with Joaquin Vanschoren on vision-language-action models for robotics. In 2026, I am finishing my PhD on generalization in deep reinforcement learning, supervised by Decebal Mocanu and Mykola Pechenizkiy. Research interests include robotics, VLAs, RL, generalization, sparse neural networks.

During my PhD, I did an internship at Sony AI in the Gran Turismo team. We published a paper on large-scale generalization, accepted as an oral at AAAI 2026. Read the paper on arXiv and see the presentation below.

Presentation of the SPARC paper with Sony AI at AAAI, Jan 2026.

In 2023, I visited the University of Alberta for 5 months as part of my PhD, joining the Intelligent Robot Learning lab of Matthew Taylor. See the talk I gave in the AI Seminar Series by Amii below.

Presenting my PhD research at the University of Alberta, August 2023.

In 2021, I finished two masters at the TU Eindhoven called:

You can find both master theses on my TU/e page. The thesis for IAM was titled: Deep Reinforcement Learning for the cooperative card game Hanabi.

Master's thesis presentation, Sept 2021.

For my bachelor of Applied Mathematics, I started my studies at Wentworth Institute of Technology, in Boston, USA. My bachelor’s thesis used graph theory and optimization for a matching assignment problem.

Postdoctoral Researcher

2026 — present

As a postdoc in the AMOR/e lab (Advanced Models through Open Research and Engineering), I work on multimodal foundation models for robotics. We aim to increase the functionality of open-source Vision-Language-Action models.

PhD Candidate

2022 — 2026

Research on dynamic sparse training in deep reinforcement learning, improving the adaptiveness, efficiency, and focus of neural networks. We test on benchmarks such as the DeepMind Control Suite, MuJoCo, and the UR5 Robotic Arm.

Master Applied Mathematics

2018 — 2021

Graduated cum laude. Thesis on multi-agent deep reinforcement learning for the cooperative card game Hanabi. We published a paper based on this work, titled “Is Vanilla Policy Gradient Overlooked? Analyzing Deep Reinforcement Learning for Hanabi”. See the paper on arXiv.

Master Science Education

2018 — 2021

Acquired the official license to teach mathematics in Dutch high schools. Taught math classes to all levels of high school students, during two half-year internships at het Maaslandcollege in Oss and het Van Maerlantlyceum in Eindhoven.

Bachelor Applied Mathematics

2014 — 2017

Studied abroad in Boston at Wentworth Institute of Technology for my first year, after which I continued at the TU Eindhoven in the Netherlands. Played varsity baseball for the Wentworth Leopards.

CV

See my CV here. Last updated: Feb 2026. Google Scholar.