At Miuul Nordic Data Science Academy and Miuul A Lab R&D, I deliver automation and AI projects from stakeholder scoping through production deployment. I also mentor bootcamp participants in Low Code Automation and Claude Code.
For RepoRadarAI, I built a model to predict GitHub repository abandonment and a RAG recommender for maintained alternatives. The model achieved ROC-AUC 0.938, and its recommendations use semantic search over repository documents.
I created DigiSchafak to schedule bootcamp WhatsApp announcements across a 40-session term, with duplicate-safe delivery and a full-term simulation that produced zero errors. I also developed n8n workflows that turn podcast episodes and recorded lectures into SEO articles.
At the Experimental Volcanology Lab at Ludwig-Maximilians-Universität, I turned particle images into a clean dataset of more than 2,400 measurements and contributed to research published in Communications Earth & Environment. My work connects careful measurement and analysis with practical automation.
