I am a machine learning scientist, a principal investigator (H.F.R.I. project ARCHNETS) and an adjunct lecturer at the
Department of Informatics, University of Athens and at the Archimedes AI unit/Athena RC.
Additionally, I am a founding member & director of
Visible Machines,
an Athens-based non-profit organisation that researches and informs on the social dimensions of Artificial Intelligence.
I obtained my PhD in computer science from
Imperial College London,
and my MEng in electrical and computer engineering from the
National Technical University of Athens.
Previously, I spent time at
Google DeepMind,
École Polytechnique Fédérale de Lausanne,
KU Leuven,
NCSR Demokritos and
Universitat Politècnica de Catalunya.
I conduct basic and applied research on several machine learning topics, particularly within
Geometric/Graph Deep Learning, where I design and theoretically analyse data-driven methodologies
for geometric and symmetric data, such as complex networks, physical systems, 3D objects and weight spaces.
Additionally, I study AI from a sociotechnical angle, with a particular focus on algorithmic auditing
and AI interpretability. My work has been published in leading conferences (NeurIPS, ICML, CVPR, ICCV, ECCV)
and journals (TPAMI), while I regularly engage in educational activities and academic service.
Scientific interests
- AI auditing - transparency, interpretability, trustworthiness
- Societal impact of AI
- Weight space learning
- Geometric deep learning
- Generative models & self-supervised learning
- ML for biology/chemistry
News:- Oct 2026
📢 Machine Learning (MSc in DSIT, UoA) starts again this Fall on Wed, Oct 7th, 2026. I also plan to supervise new MSc theses on trustworthy AI and mechanistic interpretability / weight space learning. Get in touch if interested.
- July 2026
Our workshop proposal on Neural Network Artifacts as a New Data Modality got accepted at NeurIPS 2026 (📍 Palais des Congrès, Paris, 📅 13 December 2026).
- May 2026
📄 New paper: Exposing Hidden Biases in Text-to-Image Models via Automated Prompt Search (ICML 2026).
- Nov 2025
New research grant: H.F.R.I. research project for postdoctoral researchers (Principal Investigator): “NNs as Architects (ARCHNETS)", start-end date: April 2026- April 2028
- Oct 2025
New graduate course: Machine Learning at the MSc Data Science and Information Technologies, Department of Informatics, UoA
- July 2025
Visible Machines, our non-profit organisation has been launched.
- Mar 2025
📢 Weight space learning workshop accepted papers announced.
- Dec 2024
Our workshop proposal on Neural Network Weights as a New Data Modality got accepted at ICLR 2025 (Singapore, 27 April 2025).
- Sep 2024
📄 New paper: Scale Equivariant Graph Metanetworks (NeurIPS 2024, oral).
- Apr 2024
📄 New paper: Bridging Mini-Batch and Asymptotic Analysis in Contrastive Learning: From InfoNCE to Kernel-Based Losses (ICML 2024).