I build digital products.

Project Description Claimwise is a web application that helps users better understand insurance documents. Users can upload an insurance policy as a PDF and ask questions such as “Is this claim covered?” or “How much is my deductible?” The application finds relevant passages in the document and uses AI to generate an answer. Each answer includes the document name, page number, and a text excerpt, allowing users to verify the information directly in the original document. A machine learning model also assesses selected claims data and flags cases for manual review. The application does not decide whether an insurer should pay a claim. Claimwise is my first AI project and a portfolio prototype built using synthetic insurance documents and training data. It demonstrates how document search, AI-generated answers, and support for claims assessment can work together in a single application.

A standard QR code quickly reaches its limits, with a maximum capacity of around 3 KB. This web project is an interactive demo developed as preliminary work for my bachelor’s thesis. It explores how mathematically layered codes, known as Multi-Layer QR codes, can significantly increase data capacity. The functional prototype aims to demonstrate how complex datasets can be transferred and managed directly through a visual medium, without relying on external servers or centralised infrastructure. I translated the mathematical principles described in research papers into a custom algorithm tailored to the prototype.

Parkoala ist ein intelligenter Parkplatzmanager, mit dem sich die Ein- und Ausgaben der verfügbaren Parkplätze im Unternehmen über das Dashboard steuern lassen. Mitarbeiter können ihre Parkplätze über eine App buchen.