Welcome to my website! I am especially interested in mathematics, the theory of computer science, computational biology, and the impact of technology on society. I am currently a student in the Mathematics of Random Systems Doctoral Training programme, run jointly between Oxford and Imperial. Before, I studied Computational Biology at the Department for Applied Mathematics and Theoretical Physics (DAMTP) in Cambridge, as well as mathematics and computer science at the University of Oxford and ETH Zurich, spending some time as visiting student at UC Berkeley. My doctoral studies are supported through a Rhodes Scholarship.
Mathematics & Theoretical Computer Science x Computational Biology
The Moran process is an stochastic process describing how a ‘mutation’ spreads through a population. When working with this model, one faces two main problems:
- The Moran process accounts only for a single type of “mutation”, making the model useless when applying it to misinformation spreading (there are usually multiple competing ideas) and cancer evolution (there are realistically multiple (driver) mutations).
- The (spatial) Moran process is computationally complex so that the running time of deterministic algorithms grows exponentially in the network size (misinformation spreading application) or tissue size (biological application).
As part of my Masters thesis, we developed an efficient Markov chain Monte Carlo algorithm for the spread of multiple competing mutation types (or ideas) through a spatial population of any topology, solving problem (1). The arbitrary population structure (e.g. a biological tissue structure or social network) is specified in form of a graph. The algorithm is a fully polynomial randomised approximation scheme, hence it works efficiently even for large populations through our effective use of randomness, solving problem (2). The key step is a lower bound on the fixation probability, which is polynomial in the graph size. Submitted; preprint here.
Randomized controlled trials (RCTs) are the gold standard for evaluating a treatment’s effectiveness in medicine and have spread to many other disciplines like behavioral economics. However, state-of-the-art methods are either biased or inefficient in the presence of spillover effects. We (1) formulate a graph-based model that accounts for network interference and (2) design a novel statistical estimator that is unbiased, even in the presence of interference. Techniques used range from probability to graph theory. [arXiv]
Randomized matchings become increasingly important in behavioral economics (randomized experiments) and medical trials (randomized clinical trials). I gave a talk about this topic at the ETH Center for Law and Economics.
Presentation on Linear Stability Analysis.
Presentation of the complexity-theoretic paper Playing Dominoes Is Hard, Except by Yourself (Demaine et al.) at the Presenting Theoretical Computer Science Seminar at ETH Zurich.
Technology & Society
- I co-founded the seminar Europe’s Digital Future – Internet politics. Cybersecurity. Data rights. We invite speakers from academic scholars to policymakers to explore how digital technologies transform society, funded by the German Academic Scholarship Foundation. Talks are on a bi-weekly basis. Feel free to contact me with any inquiries and refer to digital-ftr.eu for upcoming talks.
- Having worked in Prof. Bechtold’s intellectual property group at the ETH Center for Law and Economics, I am happy to have supported empirical research based on game theory and behavioral experiments in the interdisciplinary field of intellectual property, Internet, privacy, telecommunication and law & technology. Example projects I supported are on explainable AI (XAI) and so-called moral rights: Do particular legal rights granted by copyright law (moral rights) generate incentives for artists to create excellent art?
- Panel Discussion with Germany’s Federal President Gauck #DE2036 – Wie soll es aussehen, dieses Land? Deutschland in 20 Jahren on Germany in 2036, heading the team on European politics
- I worked in the German Parliament for MP and Federal Minister ret. Dr. Ramsauer, Committee Chair on Economic Cooperation and Development
- I built a drone-detection system. My algorithm is primarily a computer vision algorithm with some sound analysis. The system can perform real-time drone detection and classification in 3d using just simple webcams. The user gets information such as 3d position and velocity. It is now possible to point a laser pointer automatically in real-time at the drone.
- Prior projects include basic shape recognition algorithms and projects at the intersection of computer vision and robotics.
See the page ‘personal’ for personal coding projects as well as my GitHub account for public code. I have also worked in a django framework for experimental studies, which is code for the ETH Center for Law and Economics and thus in private repositories.
- Rhodes scholarship
- Intel International Science and Engineering Fair
- Best of Category Award in Robotics and Intelligent Machines
- First Grand Award
- European Union Contest for Young Scientists:
- Second Award
- EIROforum special prize
- Jugend forscht (Europe’s biggest youth science competition):
- Chancellor’s Award
- Federal winner (German: Bundessieger)
- Europe Award by DFG (German Research Foundation) and European Commission
- Presentation to Chancellor Angela Merkel (incl. Video)
- Presenting for the Federal Ministry of Education and Research at CeBIT