Biomedical engineer turned founder & AI engineer
Ghent, Belgium
I'm the technical co-founder of CovaCtrl, a Ghent-based startup reinventing how enterprises understand, manage, and assure internal control. We built an agentic AI platform that continuously monitors internal controls, replacing manual testing with a real-time, evidence-based view of risk and control effectiveness. It's built for risk & control teams at enterprises. The stack: React, TypeScript, and a modern cloud infrastructure.
View CovaCtrl on LinkedIn
Built an AI agent that automatically cleans and maps unstructured client data, like spreadsheets and raw exports, so it can be ingested into ML models. Also built an interactive agent that lets clients access Toqua's services in natural language.
A day trading system for US listed tickers. I developed ML and deep learning models that predict next-day price movement from historical correlations. Based on those predictions, the system sizes and places orders through a paper-trading broker that simulates real market conditions.
I built a sentiment analysis engine that scavenges the web for company mentions (articles, social posts, reviews) to surface how a brand is actually perceived online. It scores sentiment, tracks competitors and recurring topics, and visualizes it all in an interactive dashboard.
For my thesis, I worked with Ghent University’s MEDISIP research group on a next-generation PET scanner that has no built-in CT component, meaning there’s no straightforward way to correct the scans for how the body absorbs radiation. I trained a deep learning model to predict that correction directly from the PET data itself, then tested it against real patient scans. It proved accurate enough to replace the missing CT step, without the added radiation dose or hardware cost.
Tap a figure to enlarge