In my Bachelor Thesis i explored the novel technique of Variational Graph Autoencoders to generate realistic, synthetic traffic scenes. In addition I also clustered the resulting latent space to make complex traffic scenes comparable.

As a follow-up to my Thesis and while working at the FZI We wrote and published this paper about using Graph Neural Networks and Contrastive learning to generate a metric for comparing traffic scenes without expensive graph matching. <Preprint>

Comparing and Communicating Coffee grind sizes and distributions between different grinders is difficult. Professional lab tools exist but access to them is expensive. This simple website provides a free and reliable solution. <Analyzer>

As part of the Einstein Motorsport team I developed, simulated and build the rear aerodynamic components. At Formula Student Austria I defended our aero design sucessfully.

Starting as a hobby project and later on financially supported by the Baden-Würtemberg-Stiftung We developed an Electric Longboard with an innovative weigth sensing control scheme, making for a more intuitive and safe ride. Participating at Jugend Forscht we won the first place in the regional competition and won a sponsored internship at KIT at the state competition.