Crafted an AI-powered platform that enables Swiss citizens to draft popular initiatives through real-time human-AI collaboration, transforming direct democratic participation with cutting-edge language technology.

Done as a Master's thesis in collaboration with the DDIS Lab at the University Zurich.

Developed an open-source mobile platform that simplifies user studies by enabling researchers to deliver and monitor multimedia content in real-time, with a focus on news diversity and AI-driven content recommendations. Used in academic research to study how content diversity and exposure to minority voices affect user engagement.

Done as a Master's project in collaboration with the DDIS Lab at the University of Zurich.

Developed a semantic search backend for Switzerland's largest employer, leveraging large language models and vector embeddings to match job seekers with relevant positions. The system understands the meaning behind job descriptions and search queries, enabling more intelligent job recommendations beyond keyword matching.

Redesigned the website of a wealth investment company.

Website

Developed a web application to recommend the right statistical tool for sampling datapoints. The application was developed together with a Ph.D. student at the Zurich People and Computing lab.

VirtualWardrobe is a cross-platform application I created for my bachelor thesis. The premise is simple, make your wardrobe portable by digitally carrying your pictures of your clothing with you.

A custom machine learning model processed the pictures to crop and categorize the clothes. Embedding the machine learning model ensured that the pictures stay private and never leave your own devices (iOS, iPadOS, macOS) or private cloud (iCloud).

This project was done in a cooperation between my German university and a big german client. The goal was to predict the day the customer is most likely to purchase an item. The development of an artificial intelligence to solve this task was documented in a scientific thesis

Padessi is the outcome of an internal company hackathon. Our idea was a mobile application that showed employees of the company free parking spaces.

A Raspberry Pi equipped with a camera periodically took pictures of the company parking lot. An artificial intelligence on the Pi analyzed the images and reported the free parking spaces to a server. Our team won third place out of more than 30 teams.