Open Source
I am an advocate for open science. Through my academic journey, I directly contribute datasets and code around motoneuron electrophysiology, cross-session/task high-density surface EMG, motor unit physiology and activity, and neuromuscular modelling.
I also push open-science initiatives through my dual role as CSO of Yneuro and Honorary Research Officer at Imperial College London: developing and maintaining code libraries, organizing open competitions, and building bridges between academia and industry.
I strongly believe that science progresses when academia and industry work hand in hand around shared tools, open benchmarks, and reusable resources on a large scale.
Open Challenges
I co-lead with Bruno Aristimunha the organization of the EEG/EMG Foundation Challenge 2026. The competition connects academic labs, open-source infrastructure, and industry research teams around EEG/EMG foundation models and domain shifts across subjects, sessions, emotions, context, and hardware. Yneuro is the lead organizer in collaboration with Inria, UCSD, and Meta, backed by AWS, with five tracks:
Infrastructure
Supporting open-source libraries and benchmark platforms that make shared scientific progress easier to reproduce, maintain, and scale.
Braindecode
Building the Yneuro x Inria Paris-Saclay collaboration to contribute to, sponsor, and maintain Braindecode.
Codabench
Commissioning AWS and Deloitte to support the Codabench team at Paris-Saclay to scale the competition platform.
MNE-EMG
Initiating, with Pranav Mamidanna, a central Python package for EMG signal processing and analysis within the MNE-Python ecosystem.
Python Packages
Releasing reusable code around EMG, motoneurons, motor units, simulation, and neuromuscular modelling.
MUniverse
Simulation and benchmarking suite for motor unit decomposition, spanning synthetic, hybrid, and experimental EMG with ground-truth motor unit spikes.
Motoneuron-Driven Neuromuscular Model
Subject-specific neuromuscular modelling with motor unit resolution.
NeuroMotion Simulator
Dynamic, motion-driven EMG signal generation with neuromechanical and deep learning models.
Data Augmentation for Motoneuron Pool Reconstruction
EMG-based motoneuron pool reconstruction with data augmentation.
Data Release
Making datasets reusable for motor unit physiology, high-density EMG, neuromuscular simulation, and cross-species motoneuron modelling.
High-Density EMG Dataset II
Signals and manually edited spike trains from 2 muscles, 16 participants, 8 contractions levels, 1 session, with 2000+ identified motor units.
High-Density EMG Dataset I
Signals and manually edited spike trains from 1 muscle, 6 participants, 2 contraction levels, 1 session, with 500+ identified motor units.
Medical Imaging & MSK Model Dataset
Segmented anatomical volumes and subject-specific musculoskeletal models for neuromuscular simulations.
Cat & Rat Motoneuron Electrophysiology Dataset
Experimental motoneuron electrophysiology data for cross-species modelling.