Zaccharie Ramzi

Zaccharie Ramzi

Postdoc Researcher in Deep Learning




I am a Postdoc Researcher supervised by Gabriel Peyré at ENS Ulm (CNRS). I obtained my PhD in deep learning for MRI reconstruction under the supervision of Philippe Ciuciu and Jean-Luc Starck at Parietal (Inria), NeuroSpin and Cosmostat (CEA).

Download my résumé.

Download my PhD thesis manuscript.

  • Artificial Intelligence
  • Computer Vision
  • Deep Learning
  • AI for Healthcare
  • Open Source
  • PhD in Artificial Intelligence, 2022

    CEA & Inria

  • MSc in Mathematics, Vision and Machine Learning (MVA), 2017

    Ecole Normale Superieure Paris-Saclay

  • Engineering Diploma, 2017

    Telecom ParisTech


Postdoc Researcher in Deep Learning
Apr 2022 – Present Paris, France
  • Working on the understanding and improvement of Deep Equilibrium Models
  • Working on Benchopt, a benchmarking framework for optimization algorithms
PhD Student
CEA & Inria
Feb 2019 – Feb 2022 Paris, France
  • Designed and implemented new models for MRI reconstruction which allowed to me to secure the 2nd spot in the fastMRI 2020 reconstruction challenge organized by Facebook and NYU. This was featured in 2 articles in the specialized press (CEA mag' and Contact).
  • Used the public HPC Jean Zay to train neural networks in a distributed fashion with up to 8 nodes totalling 32 GPUs while working on 1Tb of MRI data. Co-created a user’s collaborative documentation.
  • Co-founded a lecture group focused on Deep Learning for students at NeuroSpin.
  • Analyzed and benchmarked the state-of-the-art in Deep Learning for MRI reconstruction.
  • Published 18 papers/abstracts in international peer-reviewed conferences/workshops/journals, among which IEEE TMI and ICLR (spotlight).
  • Co-supervised 2 interns: Sophie Starck on the topic of GANs for reconstruction, and Kevin Michalewicz on the follow-up of my work on Learnlets.
  • Peer-reviewed 19 submissions for scientific conferences and journals.
Data Scientist
Oct 2017 – Dec 2018 Berlin, Germany
  • Built data pipelines for smartphone and wearable data.
  • Designed, implemented, deployed and maintained ad-hoc machine learning models for human activity detection.
  • Presented research results to technical and non-technical teams.
Research Intern
Apr 2017 – Aug 2017 Paris, France
  • Benchmarked and implemented deep learning algorithms for epilepsy detection in EEG signals.
Data engineer and Data science Intern
Mar 2016 – Sep 2016 New-York, USA
  • Implemented and maintained data pipelines for genomic literature meta-analysis.
  • Developed a blood hormone level model.
Data science Intern
Aug 2015 – Feb 2016 Paris, France
  • Developed the full stack of web applications.
  • Performed data analysis on the use of these web applications, as well as on social media data.

Recent Publications

Quickly discover relevant content by filtering publications.
Benchopt: Reproducible, efficient and collaborative optimization benchmarks
Numerical validation is at the core of machine learning research as it allows to assess the actual impact of new methods, and to …


I gave the following courses: