Research Timeline

Research Assistant at Heidelberg Institute of Global Health (HIGH)

Benchmarked time-encoding strategies (sinusoidal, learned embeddings, text serialization, LLM encoding) for mortality risk prediction on UK Biobank (n=124,158).

Master's Thesis Researcher at German Cancer Research Center (DKFZ)

Thesis on top of Elyas Heidari's Segger model: an interpretability framework for its attention and embeddings, a study of the Aligned Segger extension, and a benchmark across 8 cancer datasets.

Research Assistant at Heidelberg Universität

Developed a generative model for discrete data by projecting high-dimensional flow matching (Generative Assignment Flows).

Research Assistant at CAMS & PUMC

Led clustering analysis identifying phenotypes of COPD patients from a nationwide cohort (n>10,000); co-first author, published in Engineering (2025).

Research Assistant at CAMS & PUMC

First-author discrete choice experiment on COVID-19 vaccine preferences among 12,000 adults in China; contributed to 5+ peer-reviewed publications.