Data Scientist & AI Researcher
I am a data scientist working at the intersection of natural language processing and law. My research focuses on building and evaluating models for Legal Named Entity Recognition and Relation Extraction, as well as constructing citation networks and structured datasets for German legal texts. I am particularly interested in how domain-adapted language models can support reliable, privacy-compliant processing of court decisions and legal documents.
Beyond model development, I work on turning research prototypes into usable systems, including anonymization pipelines, similarity search, and graph-based analysis of legal case relationships. My goal is to combine careful data modeling with robust machine learning to make complex legal texts more accessible and analyzable at scale.
Data Scientist / KI-Experte
- Developed a citation network of German legal cases using Neo4j, allowing for efficient cross-referencing and analysis of case relationships.
- Enhanced the previous Legal NER LLM to automate the anonymization of German legal documents, ensuring compliance with privacy regulations.
- Implemented a legal text similarity search as a foundation for a RAG-based system to support more accurate, context-aware legal research.
Data Science Researcher
- Fine-tuned a German BERT model on the legal dataset, achieving an F1 score of 99.49%, and compiled a dataset of 2944 German legal references.
- Co-developed a GDPR-compliant dataset of 44 privacy policies with 33 entity types, enhancing Named Entity Recognition (NER) and Relation Extraction (RE) capabilities for NLP models.
- Independently fine-tuned and published LLMs for NER and RE in privacy policies, achieving F1 scores of 74% and 83%, respectively.
Junior Android Developer
- Developed an Android application to help local businesses promote sales with hourly discounts, integrating Firebase for the real-time database.
- Wrote REST web services to support Android and iOS versions and built a business interface for managing discounts independently.
-
2025 – JURIX Legal Knowledge and Information SystemsSegmentation and Processing of German Court Decisions from Open Legal Data
-
2024 – OSSYM Open Search SymposiumA dataset of GDPR compliant NER for privacy policies
-
2024 – JURIX Legal Knowledge and Information SystemsChallenges and Considerations in Annotating Legal Data: A Comprehensive Overview
-
2023 – ICAIL International Conference on Artificial Intelligence and LawA Dataset of German Legal Reference Annotations
-
2023 – ICAART International Conference on Agents and Artificial IntelligenceGerman BERT Model for Legal Named Entity Recognition
-
2021 – KDIR International Joint Conference on Knowledge Discovery, Knowledge Engineering, and Knowledge ManagementExploring Semantic Similarity Between German Legal Texts and Referred Laws
-
2021 – LOD International Conference on Machine Learning, Optimization, and Data ScienceExperiments on Properties of Hidden Structures of Sparse Neural Networks