Portrait of Ippokratis Pantelidis

Ippokratis Pantelidis

MSc AI student at the University of Amsterdam

Video Foundation Models · Physical AI · Information Retrieval

I am an MSc Artificial Intelligence student at the University of Amsterdam. My thesis, in the VIS Lab under Cees Snoek, studies world models that act as action-conditioned verifiers in video games. Alongside it I work on two research projects: one examines the physical understanding of video foundation models, the other studies generative reranking in information retrieval.

Before this I studied Computer Science at the Athens University of Economics and Business, where I graduated first in my class with a 9.27/10. My bachelor thesis, supervised by Ion Androutsopoulos and John Pavlopoulos, was on diagnostic captioning: generating diagnostic text from medical images.

Publications

2026

  • How Do Video Foundation Models Encode Intuitive Physics? Probing Across Pretraining Paradigms

    Samuele Punzo*, Niccolò Caselli*, Ippokratis Pantelidis*, Francesco Massafra*, Salvatore Lo Sardo*, Mohammadreza Salehi

    NeurIPS 2026 Workshop on World Models in Physical AI (poster), 2026

    Probes whether physical plausibility is decodable from frozen video foundation model features (V-JEPA, VideoMAE, LTX-Video) using linear, MLP and attentive probes on Minimal Video Pairs and IntPhys2.

  • Mind the Gap: Promises and Pitfalls of Hierarchical Planning in LeWorldModel

    Niccolò Caselli, Francesco Massafra, Samuele Punzo, Salvatore Lo Sardo, Ippokratis Pantelidis, Sathya Kamesh Bhethanabhotla

    NeurIPS 2026 Workshop on Pretrained Agents and WM@Booth 2026, 2026

    Adds a high-level macro-action planner to LeWorldModel and shows that naive hierarchy can underperform flat planning; constraining the search to macro-actions seen in training makes hierarchy useful at longer horizons.

2025

  • AUEB NLP Group at ImageCLEFmedical Caption 2025

    Anna Chatzipapadopoulou*, Ippokratis Pantelidis*, Foivos Charalampakos*, Marina Samprovalaki*, Georgios Moschovis, Panagiotis Kaliosis, Kalliopi V. Dalakleidi, John Pavlopoulos, Ion Androutsopoulos

    CLEF 2025 Working Notes, CEUR Workshop Proceedings, Madrid, 2025

    The team's approach to the ImageCLEFmedical Caption 2025 challenge, placing 1st in Concept Detection, 1st in Explainability and 5th in Caption Prediction.

* equal contribution

Experience

  • Sep 2026 – Present

    Teaching Assistant

    University of Amsterdam · Amsterdam, The Netherlands

    Teaching assistant for Computer Vision 1 (MSc AI): weekly lab sessions and support on assignments and theory.

  • Jun 2025 – Nov 2025

    Research Intern

    Archimedes Unit, Athena Research Center · Athens, Greece

    Co-authored the AUEB NLP Group paper for ImageCLEFmedical Caption 2025 with the group's Biomedical Team (1st in Concept Detection and Explainability, 5th in Caption Prediction).

Education

  • Sep 2025 – Present

    MSc Artificial Intelligence

    University of Amsterdam · Amsterdam, The Netherlands

    Thesis in the VIS Lab on world models as action-conditioned verifiers in video games. Supervised by Cees Snoek, with daily supervision from Dheeraj Varghese and Ruthu Rooparaghunath.

    Teaching assistant for Computer Vision 1; Project AI A with TNO.

  • Oct 2021 – Aug 2025

    BSc Computer Science

    Athens University of Economics and Business · Athens, Greece

    Thesis on diagnostic captioning, supervised by Ion Androutsopoulos and John Pavlopoulos.

    Grade 9.27/10, valedictorian. Research member of the AUEB NLP Group (Biomedical Team), CS labs administrator, TechBiz 2024.

  • Jan 2025 – Jun 2025

    Erasmus+ exchange

    Stockholm University, Department of Computer and Systems Sciences · Stockholm, Sweden

    Research member of the EFRA project.

Contact

Open to PhD positions starting 2027 and research collaborations.

Email
ippokratispantelidis [at] gmail [dot] com
Location
Amsterdam, The Netherlands