About
My story so far
Vincent Pauline
PhD Student · Theoretical Foundations of Generative Models
Technical University of Munich · Helmholtz AI · MCML
I am a first-year PhD student and MCML Junior Fellow working with Prof. Dr. Stefan Bauer at TUM and Helmholtz AI. My research focuses on the mathematical foundations of generative modeling, with particular interest in extending diffusion and flow-based methods to general state spaces—including text, sequences, and multi-modal data.
I hold dual master’s degrees from CentraleSupélec and Paris-Saclay University, with expertise in probability theory, statistical inference, and deep learning.
News
Foundations of Diffusion Models in General State Spaces: A Self-Contained Introduction
with Tobias Höppe, Kirill Neklyudov, Alexander Tong, Stefan Bauer, Andrea Dittadi
A first comprehensive unified foundation of diffusion models on general state spaces, connecting discrete/continuous time and discrete/continuous data.
UNI-D²: Unified Codebase for Discrete Diffusion
with Kalyan Varma Nadimpalli (co-first), Ferdinand Kapl, Amir Mohammad Karimi-Mamaghan, Alexander Tong, Andrea Dittadi, Stefan Bauer
The first unified codebase for training large diffusion language models, enabling reproducible research in discrete diffusion.
Research Interests
Generative Modeling Mathematical foundations of diffusion models and flow-based methods
Discrete Diffusion Extending continuous diffusion theory to text, sequences, and categorical data
Probabilistic Machine Learning Statistical inference, sampling methods, and Bayesian approaches
Applications Diffusion language models, guidance, fine-tuning, and multi-modal generation
Timeline
2025 — Present
PhD Student & MCML Junior Fellow
Helmholtz AI · Technical University of Munich
Working on mathematical foundations of generative modeling under Prof. Dr. Stefan Bauer. Research on diffusion and flow methods for general state spaces.
2023 — 2024
Deep Learning Researcher
NeuroSpin · CEA Saclay
Developed a self-supervised generative model simulating both overt and inner speech, mimicking cognitive speech development in children. Master thesis supervised by Dr. Ladislas Nalborczyk and Prof. Dr. Thomas Hueber in Prof. Dr. Stanislas Dehaene’s UNICOG lab.
2023
Data Platform Manager Intern
Owkin · Paris
Contributed to Abstra, an AI-driven federated learning platform. Supported the design of an obfuscation toolchain to protect deep learning models intellectual property.
2022
Industrialization Engineer Intern
BioSerenity · Washington DC, USA
Contributed to the industrialization of AI-assisted diagnostic medical devices (PSG), collaborating with cross-functional teams across France, the US, and China. Supported production, quality testing, and regulatory activities, including FDA Letter-to-File and 510(k) clearance.
2020 — 2024
MSc Mathematics and Computer Science & MSc Computational Neuroscience
CentraleSupélec · Paris-Saclay University
Dual master’s in mathematics and computer science (GPA: 4.11/4.33), and computational neuroscience (GPA: 4.33/4.33). Specialization in probability, statistics, and machine learning.
2018 — 2020
Classe Préparatoire
Lycée Lakanal · Paris
Intensive preparation for Grandes Écoles entrance examinations (GPA: 4.33/4.33).