About Me

I am a Dual PhD candidate at the University of Porto (Portugal) and Leiden University (The Netherlands), with a background in computer science and a specialization in Artificial Intelligence. Before returning to academia, I spent several years as a data scientist in the industry, where I came to appreciate the importance of understanding model decisions and building data-driven solutions. This experience shaped the direction of my current research.

My PhD work focuses on the intersection of computer vision and responsible AI, particularly in stress testing. Stress testing is a methodology that identifies model weaknesses by examining their decision boundaries. For my PhD, I am developing a framework that uses generative models to create data near a classifier’s decision boundary. This approach helps us better understand how models handle low-confidence images by selecting representative boundary samples (prototypes) and applying post hoc explanation techniques. My goal is to enhance model trustworthiness by exposing potential limitations to researchers before deployment and to end users after deployment. My research is supported by the AISym4Med project, which is dedicated to advancing AI technologies for healthcare, and I’m supervised by professores Luís F. Teixeira and Jan van Rijn.

Beyond my research, I am an invited assistant professor and supervise several master’s students in from Computer Science and Data Science masters. I also mentor students at Porto Business School in Data Science studies.

Interests Link to heading

• Stress Testing of Machine Learning models
• Machine Learning Trustworthiness
• Computer Vision
• Generative Models
  • Generative Adversarial Networks (GANs)
  • Diffusion Models
• Uncertainty
• Explainable AI
  • Prototypes
  • Concept based Explainable AI
  • Global Explanations