Alessandro Pasqui

Theoretical physicist focused on computational frameworks that bridge physics-based modeling and AI.

Passionate about machine learning, with a particular interest in applying differentiable modeling and optimization to understand complex physical systems.

I recently completed my PhD at PSL Université de Paris as a Marie Skłodowska-Curie Fellow, working at the Collège de France and École Normale Supérieure under the supervision of Dr. Hervé Turlier.

My doctoral research focused on AI-based methods for inverse problems in biological cell systems, developing computational frameworks that bridge physics-based modeling, machine learning, and cell biology.

Key projects include:

• ZAugNet — a self-supervised generative model for 3D bio-imaging (built in PyTorch) that uses adversarial learning and knowledge distillation to enhance axial resolution in microscopy data. (Manuscript under revision at Nature Communications)
• VertAX — a differentiable vertex model implemented in JAX for efficient forward and inverse modeling of confluent tissues, leveraging automatic differentiation and bilevel optimization to infer cellular parameters and reproduce tissue-scale behavior. (Manuscript in preparation for Nature Computational Science)

Before my PhD, I completed a Master’s Degree in Statistical Physics at Sapienza University of Rome and the Italian Institute of Technology, where I developed high-performance algorithms for shape matching and protein–receptor interaction studies. My Bachelor’s thesis, also at Sapienza, explored percolation in models with long-range interactions using analytical and numerical methods.