I am a PhD student at the Max Planck Institute for Intelligent Systems (Tübingen, Germany), and I am happy to be co-supervised by Moritz Hardt and Bob Williamson. My research investigates sociotechnical systems, focusing on the interface between Large Language Models (LLMs) and society. Specifically, I study how LLMs function within social media platforms and the broader research ecosystem, aiming to uncover both how these models reflect reality and how they actively shape it.
News
- Jul 2026 — Presenting at ICML 2026, Seoul. Poster: Thu Jul 9, 5:00 PM – 6:45 PM KST, Hall A (#4411).
- Apr 2026 — Correct looks better: Pairwise comparisons reveal accuracy rankings (Remeli & Hardt) accepted at ICML 2026. arxiv
I'll be at ICML 2026 in Seoul — happy to chat, reach me by email.
Education
- Budapest University of Technology and Economics, Computer Engineering, BSc with honors (2016-2020)
- University of Cambridge, Advanced Computer Science, MPhil with distinction (2021-2022)
- IMPRS-IS PhD Program (2022-)
I'm interested in...
- Machine Learning
- Sociotechnical systems
- Security and privacy
Current projects
- Pairwise comparison based ranking - arxiv
- ICML 2026 [2026]
- Predictability of online speech - preprint
Contact me!
Via email: mina.remeli@tuebingen.mpg.dePast projects
- driver fingerprinting (supervised by Gergely Acs)
- second place at the Students' Scientific Association in 2018
- published in 2019
- commit classification (BSc thesis, supervised by Gabor Horvath)
- first place at the Students' Scientific Association in 2019
-
attacks in a federated learning setting (supervised by Gergely
Acs)
- published in 2022
-
exploring bad minima in transfer learning (MPhil dissertation, supervised by Ferenc
Huszar)
- got awarded the prize for Highly Commended M.Phil Project