Tim Rädsch

As a PhD student, Tim focuses on benchmarking foundation models with an emphasis on large-scale and high-quality data. While the majority of the general research efforts in the AI area focus on architectures, data topics are often overlooked. The quality of test data in safety-critical applications in the end determines the real-world (e.g. clinical) applicability of an algorithm. High-quality labels are thus absolutely critical for certification and better labels directly translate to patient benefit.

Prior to his PhD studies, Tim worked on ML training data generation in the automotive domain, where he built and led a team of 25 Data Engineers. Furthermore, he is one of the main authors of the ASAM OpenLabel standard, the predominant labeling standard in the European automotive industry.

Expertise:

• Data centric AI
• Foundation Models
• Benchmarking
• Data Annotation