### Current position
I am a PreDoc researcher (PhD student) in the Research Unit Theory and Logic at TU Wien, working with Agata Ciabattoni. I am part of the project **AXAIS – Acquiring and explaining norms for AI systems**, funded by the Vienna Science and Technology Fund (WWTF).
AXAIS aims to develop logical and AI-based methods to **acquire norms from large bodies of legal and ethical texts** and to make the resulting norm-governed AI systems **transparent and explainable**.
---
### Research interests
My interests are split between machine learning theory and logic for AI:
1. **Multi-task learning and Neural Tangent Kernels (NTKs)**
I study how different tasks interact when learned jointly in deep networks. In particular, I analyze:
- how **NTK alignment** can be used to define similarity between tasks,
- how **early training dynamics** predict which tasks benefit from being learned together.:contentReference[oaicite:5]{index=5}
2. **Logical models of case-based legal reasoning**
Together with Agata Ciabattoni, I work on the **result model** for legal precedent, and on how to extend it with **Bayesian methods** to handle inconsistent or noisy case data while preserving a transparent logical structure.
3. **Norms, deontic logic, and explainable AI**
Within AXAIS, I am interested in:
- formalizing norms and permissions in logical systems,
- connecting these formalisms to **explainable machine learning** for legal and normative domains.
---
### Background
- **Master in Logic**, University of Gothenburg (focus on algorithmic probability, models of computation, and category theory; additional coursework in machine learning and NLP).
- **Previous research assistantship** in machine learning theory (multi-task learning and representation similarity) at the Data Lab, Vrije Universiteit Brussel.