Computational Scientist · Medical AI · Dynamical Systems
Developing reliable, interpretable, and dynamics-aware AI Systems for medical imaging and neural signals.
I combine medical image analysis, neural time-series modelling, Koopman methods, and scientific machine learning to study complex biomedical systems and build trustworthy AI.
About
I am a computational scientist with a Ph.D. in Computer Science and Technology and a foundation in applied mathematics. My doctoral research focused on deep-learning methods for radiological image classification and segmentation.
My current research extends this background toward the analysis and reconstruction of neural time-series data using dynamical systems, Koopman operators, Dynamic Mode Decomposition, and sequence models.
My long-term goal is to develop clinically meaningful AI systems that are not only accurate, but also interpretable, reliable, uncertainty-aware, and grounded in the dynamics of the biological processes they model.
Research Vision
My research integrates medical artificial intelligence, reliable and interpretable machine learning, neural dynamics, and the study of neural networks as dynamical systems.
This research focuses on developing deep learning and multimodal AI methods for medical image analysis, including classification, segmentation, and disease detection across MRI, fMRI, CT, X-ray, and ultrasound. The goal is to transform complex medical data into reliable, interpretable, and clinically meaningful insights that support diagnosis and healthcare decision-making.
This research aims to develop trustworthy AI systems that go beyond predictive accuracy by addressing uncertainty, calibration, robustness, generalization, and explainability. I investigate methods for understanding model predictions, quantifying uncertainty, and evaluating reliability, particularly in biomedical applications where transparency and dependable performance are essential.
This research combines dynamical systems theory, Koopman operator methods, and machine learning to understand, reconstruct, and predict complex neural activity. Using EEG and local field potential recordings, I investigate transient neural dynamics, hippocampal sharp-wave ripples, and cross-subject generalization, aiming to develop interpretable computational representations of brain activity.
This research investigates the learning process of neural networks from a dynamical systems perspective. By analyzing parameter trajectories, Hessian and Jacobian spectra, and Koopman operator representations, I study how neural networks evolve during training, how their stability changes, and how these dynamics relate to optimization and generalization. The goal is to develop a deeper mechanistic understanding of deep learning.
Publications
ICML Workshop · High-dimensional Learning Dynamics
ICML Workshop on High-dimensional Learning Dynamics, 2026.
View paper →arXiv Preprint
arXiv:2608.16569, 2026.
View paper →arXiv Preprint
arXiv:2609.12892, 2026.
View paper →Journal of Machine Learning for Modeling and Computing
Journal of Machine Learning for Modeling and Computing, 2026.
View paper →NeurIPS Workshop · Learning from Time Series for Health
NeurIPS Workshop on Learning from Time Series for Health, 2025.
View paper →Methods & Tools
Contact
I welcome discussions in medical AI, neural dynamics, interpretable machine learning, and scientific machine learning.