Welcome
Hi, I’m Fabian Jaeger. I document my learnings on Artificial Intelligence, Science, and Software Development in my blog. My interests include interpretable and reliable AI systems, Graph Machine Learning, and the intersection of Machine Learning with the sciences. If this intrigues you, feel free to explore my blog, where I share various insights and findings. (Please note, it’s still under construction.)
Background
My background is in theoretical condensed matter and quantum physics. I graduated in 2023 from ETH Zurich. I conducted my research in the Materials Theory Group under Prof. Nicola Spaldin. I joined initially in Fall 2022 to work on the behaviour of polar metals, an unusual and even contraindicated combination that leads to some exotic physics.
There as a student researcher, I performed electronic structure calculations based on density functional theory and constructed physical Hamiltonian models to demonstrate that two phenomena, the kinetic magnetoelectric effect and the non-linear Hall effect, are universal to polar metals. This work was published in Physical Review Research with Sayantika Bhowal in APS and the pre-print can be found here.
During my studies, I took on a few machine learning modules and got facinated by the field and so I decided to get a part-time job working in Data Science for a small pharma company that aims to deliver a drug-identificaton software to pharmacies.
For my Master Thesis I then continued applying AI and worked on training state-of-the art universal machine learning force fields (UFF’s) using MACE and M3GNet for two-dimensional material using a curated dataset of roughly 88'000 structures and exploring it as an alternative method to using Cluster Expansion for alloy modelling. Some of the results are published in the paper “Improving machine-learning models in materials science through large datasets”.
My Bachelor Thesis was conducted in the group of Prof. Titus Neupert and dealt with investigating the phases of floquet-driven quantum many-body critical systems in one and two-dimensions. This included a numerical simulation part as well as the analysis of an analytical solution using conformal field theory. The plots on this one I think are kinda neat!
Other Stuff
I will also leave some of my write-ups here, in case anyone can find use in them:
- Quantenmechanik I: This is a 200 page ‘summary’ in German of the content covered in the quantum mechanics I course from the Fall Semester 2019.
- Short Survey on Bayesian Neural Networks: This is a short overview and heavily inspired by Wilson et. al “Bayesian Deep Learning and a Probabilistic Perspective of Generalization”