About
Hi there 👋! I'm Andrei, I am currently a Ph.D. candidate at the International Max Planck Research School for Intelligent Systems and ELLIS, under the co-supervision of Mathias Niepert (University of Stuttgart) and Karsten Borgwardt (ETH Zürich & MPI-B). I am also a Machine Learning Researcher at Bitdefender's Theoretical Research Laboratory. I got a B.Sc. in Mathematics and an M.Sc. in Artificial Intelligence from the University of Bucharest. In the past I have also worked as a Software Engineer for NXP Semiconductors.
I've served as a reviewer for EMNLP 2022, ICML 2022/2023/2024/2026, NeurIPS 2022/2023/2024/2025. I got a Best Reviewer award for NeurIPS 2022, 2023 and 2024, and was named a Gold Reviewer at ICML 2026.
My Ph.D. research focuses on the design and development of Geometric Deep Learning methods where the robustness and trustworthiness of such models are an integral part of the research. My Master's thesis involved deep anomaly detection in text (full thesis).
News
-
I was named a Gold Reviewer at ICML 2026!
-
Two papers accepted at ICML 2026! Protein Fold Classification at Scale: Benchmarking and Pretraining was accepted as an Oral, and Learning the Neighborhood: Contrast-Free Multimodal Self-Supervised Molecular Graph Pretraining was accepted as a Poster.
-
One paper accepted at the ICML 2026 Workshop on Weight-Space Symmetries: How the Optimizer Shapes Learned Solutions in Equivariant Neural Networks.
-
Our paper, Learning (Approximately) Equivariant Networks via Constrained Optimization has been accepted as an Oral 📢 at NeurIPS 2025! I will also be presenting the paper at the NeurIPS NPGML workshop and at the COML workshop.
-
Two workshop papers accepted at NeurIPS 2025! Come talk with us about Learning the Neighborhood: Contrast-Free Multimodal Self-Supervised Molecular Graph Pretraining at the NPGML Workshop, and ChronoGraph: A Real-World Graph-Based Multivariate Time Series Dataset at the BERT2S Workshop!
-
I will be attending the 2025 edition of the East European Summer School - ping me if you'd like to have a coffee :).
-
One paper on Rewiring and Virtual Nodes at NeurIPS 2024!
-
I got a Best Reviewer award at NeurIPS 2024!
-
I attended the 2024 edition of the East European Summer School as a Teaching Assistant and an Industry Speaker! My keynote talk for Bitdefender is available here.
-
One paper on Graph Rewiring at ICLR 2024 🤩!
-
I got a Best Reviewer award at NeurIPS 2023!
-
One long paper at EMNLP 2023! ✨
-
One paper at the ICDM 2023 AI for Time Series workshop!
-
One paper at the ICML 2023 Differentiable Almost Everything workshop!
-
One short paper accepted at EMNLP 2022! 🤩
-
I became a Ph.D. candidate at the University of Stuttgart! 🎉
-
Two papers accepted (VeriDark & AnoShift) at the NeurIPS Datasets & Benchmarks track! 🤩
-
I attended the 2022 edition of the East European Summer School as a Teaching Assistant!
-
I presented our paper, DATE: Detecting Anomalies in Text via Self-Supervision of Transformers at the ICML 2021 UDL Workshop!
-
I attended the 2021 edition of the East European Summer School as a Participant and won a Best Poster award!
-
I presented our paper, DATE: Detecting Anomalies in Text via Self-Supervision of Transformers at NAACL 2021!
-
I got my Master's degree from the University of Bucharest! (Thesis)
-
Me and my team from the University of Bucharest have won the 1st prize at the ICLR 2021 Computational Geometry & Topology Challenge!
-
Me and my team have won the 1st prize at the 2020 PyTorch Summer Hackathon with our healthcare assistant, Q&Aid!
Publications
* equal contribution
2026
- Protein Fold Classification at Scale: Benchmarking and Pretraining ICML 2026The Forty-third International Conference on Machine Learning Oral
- Learning the Neighborhood: Contrast-Free Multimodal Self-Supervised Molecular Graph Pretraining ICML 2026The Forty-third International Conference on Machine Learning
- How the Optimizer Shapes Learned Solutions in Equivariant Neural Networks ICML 2026 Workshop on Weight-Space Symmetries
2025
- Learning (Approximately) Equivariant Networks via Constrained Optimization NeurIPS 2025The Thirty-ninth Annual Conference on Neural Information Processing Systems Oral
- ChronoGraph: A Real-World Graph-Based Multivariate Time Series Dataset NeurIPS 2025 BERT2S WorkshopRecent Advances in Time Series Foundation Models: Have We Reached the 'BERT Moment'? (NeurIPS 2025 Workshop) Oral
2024
- Probabilistic Graph Rewiring via Virtual Nodes NeurIPS 2024The Thirty-eighth Annual Conference on Neural Information Processing Systems
- Probabilistically Rewired Message-Passing Neural Networks ICLR 2024
2023
- AD-NLP: A Benchmark for Anomaly Detection in Natural Language Processing EMNLP 2023Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing
- Probabilistic Task-Adaptive Graph Rewiring ICML 2023 Workshop on Differentiable Almost Everything
- Time Series Anomaly Detection using Diffusion-based Models ICDM 2023 AI for Time Series Workshop
2022
- Rethinking the Authorship Verification Experimental Setups EMNLP 2022
- VeriDark: A Large-Scale Benchmark for Authorship Verification on the Dark Web NeurIPS 2022 Datasets and Benchmarks TrackProceedings of the 36th Conference on Neural Information Processing Systems, Datasets and Benchmarks Track
- AnoShift: A Distribution Shift Benchmark for Unsupervised Anomaly Detection NeurIPS 2022 Datasets and Benchmarks TrackProceedings of the 36th Conference on Neural Information Processing Systems, Datasets and Benchmarks Track
2021
- DATE: Detecting Anomalies in Text via Self-Supervision of Transformers NAACL 2021Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies
- ICLR 2021 Challenge for Computational Geometry & Topology: Design and Results arXiv preprint