Ziyad Sheebaelhamd

About Me

I am a Ph.D. student at the University of Tübingen and a part of the International Max Planck School for Intelligent Systems. I am supervised by Prof. Claire Vernade, where I am working on bridging the gap between theory and practice of applying reinforcement learning algorithms at scale.

I graduated from ETH Zürich in 2022 where I studied Robotics, Systems and Control with a focus on Reinforcement Learning and Optimal Control. Afterwards, I worked as a machine learning engineer at Merantix Momentum till 2024, where I developed and trained models at a large scale for various industry projects.

Research Interests

• Reinforcement Learning: RL Theory, Imitation Learning, Offline/Online RL.
• Generative Modeling: VLAs, Diffusion Models, World Models.

News

• [May 2026] Our new preprint “Split the Differences, Pool the Rest: Provably Efficient Multi-Objective Imitation” is now on arxiv!
• [Dec. 2025] Our new preprint “Efficient Personalization of Generative Models via Optimal Experimental Design” is now on arxiv!
• [Dec. 2025] I did an oral presentation for our paper “Quantization-Free Autoregressive Action Transformer” at EurIPS in Copenhagen! Slides can be found here.
• [Sep. 2025] Our paper “Quantization-Free Autoregressive Action Transformer” is accepted as a spotlight at NeurIPS 2025!
• [Sep. 2024] I started my Ph.D. at the University of Tübingen under the supervision of Prof. Claire Vernade!
• [Aug. 2021] Our paper “Safe Deep Reinforcement Learning for Multi-Agent Systems with Continuous Action Spaces” is accepted at ICML 2021 in the RL4RealLife workshop!

Publications

Talks / Lectures

Talks

• Quantization-Free Autoregressive Action Transformer
  Oral presentation for our Spotlight paper at EurIPS 2025
  [Slides]