Machine learning researcher / founder
I work on generative AI, synthetic data, and practical machine-learning systems for tabular and textual data. I am Co-Founder of tabularis.ai and hold a PhD in Machine Learning and Computer Science from the University of Tuebingen.
My current work sits at the intersection of synthetic data, language models, and research automation. I care about systems that are technically strong, measurable, and useful outside of benchmarks.
- Generative AI and LLMs
- Synthetic and tabular data
- AI agents and autoresearch
- Safety, DPO, and GRPO
tuetoken
A fast tokenizer backend for LLMs, benchmarked up to 30x faster than tiktoken or Hugging Face tokenizers.
Faust-1
A 1.6B-parameter German language model trained from scratch and designed to run efficiently on consumer hardware.
Multilingual Sentiment Analysis
A specialized 23-language sentiment model with more than 500,000 monthly downloads on Hugging Face.
Synthetic Tabular Data
Research and software for realistic synthetic tabular data generation, used by Google (Kaggle), AWS, and practitioners across industry and research.
Published tuetoken
Released a fast tokenizer backend for LLM workloads.
Released Faust-1
Published an efficient German language model trained from scratch.
YapBench on arXiv
Introduced a benchmark for measuring verbosity and over-generation in chatbot LLMs.
NeurIPS 2024 workshop
Co-organizing a workshop on tabular data representation learning.
I am open to research collaborations, internships, thesis supervision, and focused AI/ML consulting. For project discussions, send a short note with the problem, data, and timeline.