Leuven, Flemish Region, Belgium
As a Machine Learning Engineer at ML6, I'm passionate about building production-ready code within a dynamic team environment. My current focus is on developing cutting-edge agentic workflows, a great opportunity to engage with the latest advancements in AI. I hold a PhD in Trustworthy AI from KU Leuven, where my research explored trustworthy AI for high-stakes decision-making, and I also hold a Master's in Business Engineering with a specialization in Data Analytics and Business Applications. In general, I'm passionate about building AI systems, with a special heart for explainable AI, adversarial robustness, and fairness and bias. Always happy to discuss these topics!
Develop and deploy scalable data pipelines using agentic workflows with LLMs.
Designed explainable ML methods increasing the trust and adoption of AI-based decision systems in business settings As a researcher, I specialized in evaluating and enhancing the transparency and robustness of process-driven decision models, using business data from diverse sectors, including financial loan admissions, medical patient admissions, and building permit processes. In total, I collaborated on 12 papers, including Q1 journals (IEEE TSC, EJOR, ESWA) and two top-tier conference papers (ICPM). Besides my research, I also supervised 17 MSc data science projects (two of which were award-winning teams), collaborating on both academic and industry challenges. I also have the opportunity to present at invited talks on ethical and explainable AI to both academic and corporate audiences, with attendance of up to 50 people.
Developed a Python data science tutorial on descriptive analytics and advanced multivariate statistics, used for a course of 100+ B.Sc. students learning the fundamentals of programming with Python. The tutorial covers everything from collection types, NumPy arrays or Pandas data frames, and progressed to more advanced data science case studies.
Developed counterfactual models for ICU admission prediction, enabling clinicians to identify patient risk factors. Paper: Plausible and Feasible Counterfactuals for Predictive Process Monitoring (published in IEEE TSC)
Mitigated gender bias in Kiva’s lending recommendations, increasing overall fairness by 39.97% across six key fairness metrics. Paper: Explainability and Fairness in Machine Learning: Improve Fair End-to-end lending for Kiva (Published in IEEE SSCI)
Built customer segmentation models to identify the top 10% of high-value clients, enabling targeted sales strategies.