Sophie Ostmeier

I am a CS master's student and research assistant at Stanford University, where I work at the AIMI Center.

From 2023 to 2025 I was funded by the German Research Foundation (DFG). Before Stanford, I completed an M.D. and Dr.med. at the Technical University of Munich.

Outside the lab you will find me running, cycling, bikepacking, swimming, playing pickleball, or hiking. I follow a wide range of sports—especially tennis, triathlon, and track & field—but I can appreciate almost any activity with movement.

Email  /  Google Scholar  /  Stanford Profile  /  Github  /  LinkedIn

Research

My research focuses on how machine-learning models learn, make predictions and how we measure that, with a current emphasis on vision and/or language models. I am also curious about reinforcement learning as a path toward intelligent decision-making systems and wonder how they would do medicine.

Some recent papers (Google Scholar)

We extend the rotational positional encodings widely used in large language models to high-dimensional rotation matrices by exploiting their Lie-group structure, and we test this approach on both 2-D and 3-D vision tasks.

We present GREEN, an open-source metric that employs language models to spot and explain clinically significant errors in radiology reports, yielding expert-aligned scores, interpretable feedback, and commercial-grade performance.

We investigate evaluation metrics for medical image segmentation that account for uncertainty, small structures, and empty reference annotations.

Miscellaneous

Website Template source code.