关于
I am working in the Copilot Tuning team on customizing Copilot systems for customer-specific applications.
Focus: Scalable machine learning, deep learning for medical imaging, reliability and explainability.
Brief biography
I studied computer science at Graz Unversity of Technology, Austria. (opens in new tab) In May 2004, I joined the Intelligent Data Analysis Group at Fraunhofer FIRST in Berlin as a postdoc researcher. I was leading a project with Bayer Schering Pharma (opens in new tab) developing machine learning methods to predict properties of chemical compounds in early drug discovery.
In February 2008, I joined Microsoft Research as a member of the core algorithms team of Microsoft Advertising. My main project was the Bing Marketplace Simulator, an off-policy simulator for tuning auction marketplaces.
In 2017, I switched to medical imaging and was leading the machine learning engineering team inside of Project InnerEye. We later also developed MAIRA, an AI model to create draft radiological reports from X-rays and other medical images.
In Dec 2025, I joined the Copilot Tuning team, working on finetuning LLMs for enterprise customers.
精选内容
MAIRA-2: Grounded Radiology Report Generation
Radiology reporting is a complex task that requires detailed image understanding, integration of multiple inputs, including comparison with prior imaging, and precise language generation. This makes it ideal for the development and use of generative multimodal models. Here, we extend…