This project consisted of two interconnected subprojects. 1️⃣ Story GenerationThe first part focused on building a system capable of constructing short narratives — ranging from 30 seconds to 5 minutes — based on a user prompt. The main challenge was designing an agentic workflow that could autonomously create coherent, engaging,
After recent developments in GenAI, particularly in Vision-Language Models (VLMs), the fashion project I started 1.5 years ago to develop an outfit recommendation system seems already outdated. The project is outlined here. In early 2023, using VLMs was already an option; however, due to the insufficient maturity of available models, I chose a different approach described here. Driven by curiosity, I decided to revisit the project in a reduced form, as scalability is not a concern for this iteration.
This project is described in more detail below under my job experiences. I’m mentioning it here separately because I spent almost 2 years working on the system and developed deep expertise in self-learning AI architectures, large-scale data pipelines, and SAP and AWS-based integrations. Unfortunately, due to strict security and confidentiality
Fashion is highly subjective, making it difficult to develop a system that can universally recommend stylish and diverse outfits. Individual preferences, sartorial literacy, and personal experiences greatly influence how people perceive and categorize styles. Developing an algorithm that can navigate these nuances and deliver personalized, accurate recommendations is a complex task.
I was responsible for the entire technical concept, feasibility study, and client communication. Over the course of the project, I developed a deep learning–based system for automated image captioning. The main challenge was to combine multiple data sources — including the image itself — into a single grammatically correct German
I conducted an in-depth investigation of a newly collected EEG dataset to explore whether measurable differences exist between individuals with high and low hypnotizability. The study focused on two key aspects: spectral power and functional brain connectivity. After an extensive literature review and the application of advanced statistical methods, I