Projects
Selected research and software projects spanning antibody engineering, interpretable AI, and scientific computing.
ProteinTuneRL
ProteinTuneRL supports my work on protein and antibody sequence optimization with infilling language models and reinforcement learning. It turns protein-design workflows into a reusable framework for training, evaluation, and rapid experimentation across multiple design objectives.
🔗 GitHub
📚 Related paper: Reinforcement Learning for Antibody Sequence Infilling
📅 November 2024 – Present
ProtLib-Designer
ProtLib-Designer packages my work on combinatorial antibody-library design into a lightweight optimization toolkit. It uses deep mutational scanning-style scores and constrained optimization to generate diverse, high-quality candidate libraries for downstream screening.
🔗 PyPI · GitHub
📚 Related paper: Combinatorial Optimization of Antibody Libraries via Constrained Integer Programming
📮 Project post
📅 November 2024 – Present
Deep Symbolic Optimization
Deep Symbolic Optimization (DSO) is the research framework behind much of my work in symbolic regression and interpretable reinforcement learning. It provides a modular foundation for discovering equations and transparent control policies, and it has supported publications at ICLR, ICML, NeurIPS, AAAI, and related workshops.
- Symbolic Regression – recovering tractable mathematical expressions from data
- Symbolic Control – discovering interpretable policies for reinforcement-learning environments
🔗 GitHub
📚 Selected papers: NeurIPS 2022 · ICML 2021 · AAAI 2025
📅 January 2021 – Present
Cardiac Machine Learning
This project captures my work on reconstructing cardiac activation maps and transmembrane potentials from ECGs using deep learning. It combines simulation-driven data generation, model training, and evaluation for non-invasive cardiac imaging research.
🔗 GitHub · Cardiac Challenge
📚 Selected work: CinC 2022 paper · Dataset
📮 Write-up
📅 July 2018 – July 2021
Cardioid
Cardioid supported my work in large-scale cardiac electrophysiology and multiscale simulation. It enables high-fidelity finite-element modeling of electrophysiology, cardiac mechanics, and ECG forward problems on realistic anatomical meshes.
🔗 GitHub
📚 Related paper: Numerical approximation of the electromechanical coupling in the left ventricle with inclusion of the Purkinje network
📅 2018 – 2020
EXIFSI
EXIFSI (EXplicit coupling schemes for Incompressible Fluid-Structure Interaction) captures part of my earlier work on stable, efficient coupling schemes for incompressible fluid-structure interaction. The project focused on scalable multiphysics methods for challenging biomechanics and engineering problems.
🔗 Project Website
📚 Selected papers: CMAME 2016 · JCP 2015
📅 August 2012 – July 2016 (4-year ANR project)
📌 Funded by ANR JCJC (French National Research Agency)