cv

Curriculum Vitae.

General Information

Full Name Hamza Cherkaoui
Date of Birth 18th May 1991
Languages French, English

Experience

  • 2025 - Now
    Post doctoral position
    Télécom SudParis, Palaiseau, France
    • Collaborators Prof. Hélène Halconruy and Prof. Yohan Petetin
    • I focus my research on the collaboration in dynamic problem.
    • Fields of interest bandits, Bayesian optimization, Markov chains, Diffusion models, LLMs
  • 2022 - 2025
    Post doctoral position
    Huawei Technologies, Paris, France
    • Collaborators Dr. Igor Colin and Dr. Merwan Barlier.
    • I focus my research on the multi-agents linear bandits problems.
    • Fields of interest multi-/single agent(s) linear bandits, Bayesian optimization, optimization, LLMs.
  • 2021
    Post doctoral position
    BioMaps research team, CEA-Saclay, Orsay, France
    • I continue my research on the estimation of the Haemodynamic Response Function (HRF) from fMRI data in a pharmacological context.

Education

  • 2017–2021
    PhD
    University Paris Saclaye, Palaiseau, France
    • Research subject ‘Efficient whole brain estimation of the haemodynamic response function for TV-regularized semi-blind deconvolution of neural activity in fMRI’
    • Supervisors Dr. Philippe Ciuciu, Dr. Thomas Moreau and Dr. Claire Leroy
  • 2016–2017
    Master program, Data Sciences
    Ecole Polytechnique, Palaiseau, France
    • Cross disciplinary courses Machine Learning Theory, Kernel methods, Convex Optimization
  • 2015–2016
    Master program, Applied mathematics
    University Lille 1, Villeneuve d’Ascq, France
    • Focused on Probability and Statistic.
  • 2012–2016
    Master of Engineering
    Centrale de Lille, Villeneuve d’Ascq, France
    • Cross disciplinary courses centered around mathematics and computer science.

Open Source Projects

  • Metis
    • Designed for automated model selection and ensembling to optimize predictive performance
  • Bandpy
    • Designed to streamline research benchmarking while ensuring compliance with the GYM API
  • Carpet
    • Minimize 1d TV regularized problems with neural network.
  • HemoLearn
    • Disentangle the neurovascular coupling from the neural activation signal in a multivariate fashion for fMRI data.

Academic Interests

  • Curriculum Learning
    • Collaborative learning in linear bandits
    • Curriculum design in linear models
  • Automated Machine Learning (AutoML)
    • Mathematical foundations of model selection
    • Efficient hyperparameter optimization
    • Time series forecasting