I am a PhD Candidate at the Ubiquitous Knowledge Processing Lab at Technische Universität Darmstadt, supervised by Prof. Dr. Iryna Gurevych. I am working on Information Retrieval and Question Answering in the scientific domain. Previously, I did my Masters at Universiteit van Amsterdam in 2019 in Artificial Intelligence. I also worked on multimodal dialogue in Raquel Fernández's Dialogue Modelling Group. For more details, check out my current resume.
<aside> 👨‍💻 GitHub
</aside>
<aside> 🦋 BlueSky
</aside>
<aside> đź’Ľ LinkedIn
</aside>
<aside> 📧 Email
</aside>
✨ 2025/09/15 Our paper “The Good, the Bad and the Constructive: Automatically Measuring Peer Review’s Utility for Authors” got accepted at EMNLP 2025 Main Conference! Checkout the code on GitHub and find the dataset in HuggingFace.
Overview of the RevUtil dataset. Given a review comment, we annotate 4 aspects that are crucial for authors, namely: actionability, grounding, verifiability and helpfulness.
✨ 2025/01/23 My paper “PeerQA: A Scientific Question Answering Dataset from Peer Reviews” got accepted at NAACL 2025! Check out the paper on arxiv and find the code and data on GitHub!
PeerQA: A Scientific Question Answering Dataset from Peer Reviews
https://github.com/UKPLab/peerqa
Overview of the PeerQA dataset, including the data collection.
✨ 2024/10/09 I presented our paper “Best-of-Venom: Attacking RLHF by Injecting Poisoned Preference Data” at COLM! In case you missed it, check out the poster here, it gives a nice high-level overview of our findings.
https://drive.google.com/file/d/1gyYpdfwu_TRNHBmrHx6B32aW5EEEeiXz/view?usp=sharing
✨ 2024/07/28 New pre-print with collabs from Google. Check out our paper "Scalable and Domain-General Abstractive Proposition Segmentation”
Scalable and Domain-General Abstractive Proposition Segmentation
✨ 2024/04/09 New paper from my Internship at Google, London! Take a look at the tweet below to get a TL;DR, or check out the paper on ArXiv for all the details.
Best-of-Venom: Attacking RLHF by Injecting Poisoned Preference Data
https://twitter.com/timbmg/status/1777637803574100431
✨ 2023/05/23 UKP-SQuARE v3 has been accepted to ACL 2023 🥳 Check out the overview in the tweet below or find any details in the paper.
https://twitter.com/UKP_SQuARE/status/1658433479011364864
✨ 2023/03/31 New pre-print on UKP-SQuARE! We integrate multi-agent skills into the platform and compare several methods. Find all the details in the paper!