Hey, I’m Elias Baumann, a data and validation machine learning engineer at Kaiko. I build data pipelines for frontier LLM training, covering deduplication, quality filtering, synthetic data generation and schema validation across terabyte-scale datasets. I also develop LLM-based tools for clinical text pseudonymization, orchestrating distributed data processing with Dagster, Ray and Spark.

Before joining Kaiko, I completed my PhD in computational pathology at the University of Bern, where I developed deep learning pipelines for gigapixel pathology images and published on biomarker discovery in colorectal cancer, including several first-author studies. I try to develop innovative solutions for problems, where the fundamental understanding of the respective domain is crucial, and therefore also love to learn everything about the problem domain.

I am currently based in Bern, Switzerland but am working in the Zürich office of Kaiko.