I am currently working as a senior AI/ML engineer at Logitech . Previously, I was working in the research team at the Swiss Data Science Center , conducting my doctoral studies in computational statistics at ETH Zurich , and studying at TU Munich .
If you want to chat, you can reach me here:
Code
You can find my software on GitHub . Some recent projects are:
blaxbird : A PyTorch Lightning-like framework for training neural networks with Flax/NNX
sbijax : A library for neural simulation-based inference workflows in Python
ebm : Training energy-based models with noise-contrastive estimation in Python/Flax
Generative modelling, (causal) representation learning, semi-supervised learning,
time series models, signal processing, numerical computing.
Recent work
High resolution seismic waveform generation using denoising diffusion ( arXiv:2410.19343 ): A bespoke diffusion model trained from scratch on real earthquakes, i.e., audio signals, to synthesize realistic, high-frequency seismic waveforms. Keywords: denoising diffusion, score matching, signal processing
Causal posterior estimation ( arXiv:2505.21468 ): A new method for simulation-based inference that exploits the conditional independence structure of the posterior programs and encodes it into a new NN architecture. Keywords: simulation-based inference, flow matching, causality
Simulated-Annealing ABC with multiple summary statistics ( arXiv:2505.23261 ): We propose a novel ABC method that works well on high-dimensional parameter and data spaces using Simulated Annealing. Keywords: ABC, MCMC
Simulation-based Inference with the Python Package sbijax ( arXiv:2409.19435 ): A Python package implementing simulation-based inference and ABC methods in JAX. Keywords: simulation-based inference, ABC, Python, JAX
More to come soon...
Currently reading
Erich Fromm, Fear of Freedom
Erich Fromm, The Sane Society
Hartmut Rosa, Resonance
Francois Le Gall, Brownian Motion, Martingales, and Stochastic Calculus