🎓 Graduate in Computational Linguistics from the University of Zurich
🤖 Working at the intersection of NLP, Large Language Models, and Accessibility
I specialize in fine-tuning and evaluating large language models, with a particular focus on making language technologies more accessible, robust, and meaningful.
📄 Published at EACL 2026
DETECT is the first German-specific metric for holistic Automatic Text Simplification (ATS) evaluation, covering:
- ✂️ Simplicity
- 🧩 Meaning preservation
- ✨ Fluency
It is trained entirely on synthetic LLM responses, enabling scalable and consistent evaluation.
🔗 Repository:
👉 https://github.com/ZurichNLP/DETECT/
📌 Korobeynikova, Maria, et al. “DETECT: Determining Ease and Textual Clarity of German Text Simplifications.” Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics, 2026.
I’m a leading contributor to the first open repository dedicated to fine-tuning large language models for social science research.
🔗 Practitioner Guide:
👉 https://github.com/DigDemLab/OpenSource-LLM-Practitioner-Guide
📌 Alizadeh, Meysam, et al. “Open-Source LLMs for Text Annotation: A Practical Guide for Model Setting and Fine-Tuning”. Journal of Computational Social Science, vol. 8, no. 1, 2025, p. 17.
🧑💻 Natural Language Processing
- Text simplification & evaluation
- Prompting & synthetic data generation
- Model benchmarking & metrics
🧠 Large Language Models
- Fine-tuning (instruction & task-specific)
- Evaluation pipelines
- Error analysis & interpretability
♿ Accessibility-Focused NLP
- Plain language & readability
- User-centered evaluation
- Inclusive model design
🧰 Tools & Tech
- 🐍 Python
- 🔥 PyTorch
- 🤗 Hugging Face
- 📊 Evaluation frameworks
- Evaluation of LLMs beyond accuracy
- Accessibility & inclusive NLP
- Synthetic data for model training and evaluation
- Transparent and reproducible ML for the social sciences
If you’re interested in LLM evaluation, accessibility, or open NLP for social impact, feel free to reach out or explore the repos above ✨