Me, sitting at our FabLab looking into a switched-off TV screen.
At the time writing (2025-01-27), I am finishing my degree in Computer and Systems Engineering (M.Sc.) at the Ilmenau University of Technology. In my studies, I focus on mathematical modelling, data analysis, and machine learning. In particular, my interest lies in dynamical systems and bayesian statistics.
Schematic of a variational auto-encoder (VAE). A trained VAE can encode some vector x (resembling an image, soundwave etc.) as a vector z in a lower-dimensional space (called embedding space) with nice properties. This vector could then be reconstructed (decoded) to retrieve a vector similar to x again. One application of this is generating a new vector looking similar to other vectors x by sampling some random z and decoding that. This approach could be used for generating images, music pieces and in theory everything else that you could write down as a list of numbers.
In my bachelor thesis, I investigated the generation of sequential data using a variational auto-encoder and long short-term memory (LSTM) networks. In my current degree program I am continuing following these interests. An example project I finished recently, was a little research seminar about LSTM networks for modelling non-linear systems and how to train an LSTM so that it is a stable system itself. I basically looked into a recent paper about it, wrote a report and explained the proof in a talk. You can find that material here.
Schematic of an LSTM network. Adapted with changes from this blog article. An LSTM is a function approximator containing an internal state. It therefore has the capability of modelling (simulating, predicting) dynamical systems, or in other words processes that are moving in some sense. This could be anything from robots to economies.
Apart from that, I am active in our local hackspace and the KuKo e.V.. There, I organize cultural events to bring together people from different social bubbles aiming to support exchange of opinions an knowledge and to encourage the joy of creating things. An example for that are the kickelhack, I initiated in May 2024 in collaboration with the local startup service and our FabLab.
Bringing people together at music events and hackathons.
Another one are several electronic music events set up together with the local techno collective MFK. Formative experiences to do these things were my exchange semester at Aalto University in Helsinki area, where I was involved in the electronic music collective Entropy ry and several conferences in connection to the Chaos Computer Club.
