What do I do ?


Brain-computer interfaces, fundamental neuroscience

Signal processing

Physiological singals analyses (EEG, EOG, ECG)

Data sciences

Statistics, algorithms

My skills


  • Theoretical knowledge
  • Brain-computer interfaces


  • Python
  • R
  • Matlab
  • Java
  • C/C++

Machine Learning

  • Theoretical knowledge
  • scikit-learn/Scipy/pinguin
  • Pandas/Seaborn

Signal processing

  • Theoretical knowledge
  • Physiological signals
  • MNE

Education and work experiences

  • Education
  • Work experience
  • Associative
Key words :
  • Machine Learning
  • Research
  • Signal processing
  • Neurosciences
Show all
Machine Learning Research Signal processing Neurosciences

PhD Student in computer sciences

INRIA Bordeaux Sud-Ouest, POTIOC Team – Bordeaux, France

Oct. 2018 Actuel

Supervised by Fabien Lotte and Camille Jeunet

I am part of the ERC ERC Starting Grant: BrainConquest which aims to create a new generation of BCIs which ensure that users learn how to successfully encode commands.

More precisely, my objective is to create a statistical model of prediction of Brain-computer interface (BCI) users performances based on:

  • Personnality of BCI users.
  • Neurophysiological characteristics.
  • Feedback characteristics.

Teaching at the Engineering school Arts et Métiers of Bordeaux MATHS-INFOS, 64h/years (2 years)

  • Object-oriented programming (python).
  • Project management: Development of a PyQt5 graphics application.
  • Signal processing (python).


Machine Learning Research

University degree in Big Data and statistics

Graduate School of Cognitive EngineeringENSC, Bordeaux, France

2020 144 hours of training

Lien vers la formation

Research Neurosciences

International M.Sc in Bioengineering and Innovation in Neurosciences

BME Paris, PSL University, Paris-Descartes University and Arts et Métiers ParisTech. Paris, France

2016-2018 M1 and M2

Link to the formation

Research Neurosciences Signal processing

Internship in Neurosciences

INSERM,Systems Neuroscience Institute. Marseille, France

March 2017 - June 2017 Insternship Master 1

Supervised by Viktor Jirsa et Space Petkoski

I worked on The Virtual Brain Project. This project is developing a comprehensive framework for modelling the brain using computational methods built on bioengineering principles.

  • Analyzing the Epileptogenicity Index (EI) using The Virtual Brain (TVB), a computational platform that allows simulating biologically realistic large-scale brain network dynamics
  • Implementation (python) of an algorithm computing the EI.
  • Using this algorithm to test the accuracy of the EI
Signal processing

Master of Electronics from a Graduate School of Engineering

ENSEIRB-MATMECA. Bordeaux, France

2012 - 2015 3 years

Link to the formation

Signal processing Research

Internship in Research and Development

IOptima Ltd. Tel Aviv, Israël

March 2015 - Sept 2015 6 months

Internship in a startup creating a laser for glaucoma operation.

  • Product management.
  • Development of a minimally invasive ophthalmic surgical tool for the treatment of glaucoma.
Signal processing Research

Master’s project

ENSEIRB-MATMECA, Bordeaux, France

Oct 2014 - Jan 2015 4 months

In collaboration with the Institute of Cardiac Rhythm and modelling, Haut-Lévèque hospital, Pessac, FR

3D visualization of cardiac catheters by variating electric and magnetic fields.



Saft Industry. Bordeaux, France

June 2014 - Sept 2014 4 months

Worked with the team developing the lithium-ion batteries for the Airbus A350 (improvement of the electronic board reinforcing system security).

Classes Préparatoires aux Grandes Ecoles

College preparatory classes specializing in mathematics and physics

Lycée Saint Louis. Paris, France

2010 - 2012 2 years

My research works

Brain computer interfaces (BCIs) are communication and control tools that enable their users to interact with computer by using brain activity alone (which is measured, most of the time, using electroencephalography - EEG). A prominent type of BCI is mental task (MT) based BCIs, that translate modifications in brain activity induced by MTs performed by the user (e.g., imagination of movements, mental calculation or mental rotation of an object among others) into control commands for a computer. Using an MT-BCI requires dedicated training. Indeed, the user has to generate stable and distinct brain signals for each task otherwise they will not be able to control the system. Indeed, the system will not be able to recognize which task the user is performing. Producing such brain signals is a skill to be acquired and mastered and the more the user practices the better he/she will get at it. The objective of my PhD project is to contribute to the understanding of BCI user training by first doing an experimental study of learning by participating in the CYBATHLON competition. We proposed and evaluated the design of a multi-class MT-based BCI for longitudinal training of a tetraplegic user with a newly designed machine learning pipeline based on adaptive Riemannian classifiers. Using a newly proposed BCI user learning metric, we could show that our user learned to improve his BCI control by producing EEG signals matching increasingly more the BCI classifier training data distribution, rather than by improving his EEG class discrimination. In addition, this study revealed the difficulty of setting up a reliable protocol dedicated to a long term BCI training. The second part of this work is dedicated to the understanding of MT-BCI performances using predictive computational models. We proposed various computational models of BCI user training that could predict the performances of various BCI users over training time, based on BCI systems component. As a BCI is a communication system between a user and a machine such components were related to the user-profile related characteristics but also factors extracted from machine-learning algorithms used to build the system classifier. Our results suggested that is was possible to predict BCI performances using neurophysiological characteristics of a user but also neurophysiological characteristics combined with stable characteristics (i.e., traits) or the user. In addition, our studies revealed that studying features extracted from data-driven methods could be interesting to better understand why some subjects have difficulties controlling a BCI. Indeed, reliable models of BCI performances were revealed using such features.

Card image cap
The user

Models using the user characteristics.

Card image cap
The machine

Models using the "machine" characteristics.

Card image cap
The feedback

Models using the feedback characteristics.

Card image cap

Long-term BCI training of a Tetraplegic User