Towards a Philosophy of Science in the Age of AI I feel that — at a time when data technologies, and generative AI in particular, are about to dramatically change our lives and societies — the moment has come for a philosophical reflection on data and science. This text is my attempt at such a philosophy. Philosophy is concerned with completion, with the recollection and unification of parts that have been separated. While my attempt here is somewhat extensive for a blog arti

Description: In collaboration with HSLU and geoimpact, I developed a machine learning framework that estimates the real-world impact of building retrofitting measures — such as insulation and window replacement — on heat consumption. By combining predictive modeling with causal inference, the system quantifies the true energy savings achieved across diverse building types and regions. This scalable, data-driven approach enables policymakers and energy planners to track retrof

Relationships in the Emergence of the Major Ideas in Art, Science, Philosophy, and Spirituality In this article, I explore how major ideas in art, science, philosophy, and spirituality have emerged and influenced each other throughout history. Inspired by Carl Jung’s theories of the collective unconscious, I wanted to see if patterns in intellectual history could reveal something about the hidden structures shaping human thought. By analyzing historical events and their relat

Project Overview Description: For Demokratis , I helped to make Legislative Consultation PDFs more accessible and interactive by parsing them into a structured data form using AI. I employed an evaluation driven development approach, that helps to navigate the landscape of innumerable possible parsing solutions. Methods: Evaluation driven Generative AI : Constructing numeric and visual evaluation metrics using Python’s difflib and tracking experiment code and results in MLflo

Description: At geoimpact, I developed a lead generator that suggests promising buildings for the sale of a renewable energy product based on a list of past sales of the product. Methods: Thompson sampling:  This is a method for addressing the exploration- exploitation tradeoff in contextual bandit problems, where we want to maximise a certain reward (i.e., contacting the most promising building owners) and at the same time to continuously improve our model predicting the rew

Description: In my Master’s thesis at ETH Zurich, I used statistical signal processing methods for separating positional eye movement measurements into different types (saccades, smooth pursuit, and fixation eye movements). I developed a novel approach to precessing eye movement signals based on estimating signals in a mechanistic physiological model of the eye muscles. Apart from signal separation, the framework is also able to estimate the neural inputs into the eye muscles