I’m a machine learning engineer and data scientist at the FHNW Institute for Data Science in Switzerland.
I lead applied research across Swiss German ASR, multimodal modeling, and tabular ML, with deployments in production settings.
My current focus includes the research and evaluation of RAG systems and LLMs for diverse tasks, from improving ASR outputs through retrieval-augmented correction to benchmarking large language models on real-world applications. I design and implement retrieval pipelines, evaluate model hallucinations, and optimize prompting strategies for production use.
Additionally, I work on fine-tuning OpenAI’s Whisper model for Swiss German customers, open-sourcing datasets on Hugging Face, creating open-source models for space weather, and competing in the Numerai Tournament.
Before FHNW, I built data processing, classification, and NLP systems at a Swiss tech startup, working with Python and Scala.