Tübingen, Baden-Württemberg, Germany
Research assistant at the chair of Prof. Dr. Thomas Schön • First to apply regression with DNNs to electrolyte concentration prediction from electrocardiograms • Trained model on new large-scale dataset with over 290k datapoints using SLURM based GPU cluster • Compared to alternative models such as classification and ordinal regression, highlighting trade-offs • Extended practical usefulness by adding uncertainty estimation through probabilistic regression • Documented results in a research paper submitted to ML4H conference 2022
• Designed low-budget testing framework for Inertial Measurement Units enabling rapid and cost-effective prototype testing, including intuitive front-end interface • Collaborated in 4-student team on software development and implementation • Implementation with RaspberryPi, Python and the Flask library to fulfill cost, function and schedule requirements • Demonstrated real-world usage by collecting data and evaluated prototypes using the novel framework
• Gained first software development team experience with adherence to industry standards such as software patterns and SCRUM • Demonstrated learned content by implementation of a concept database for file storage in Java Enterprise, JAX-RS