Xavier Coubez, Developer in Aix-les-Bains, France
Xavier is available for hire
Hire Xavier

Xavier Coubez

Verified Expert  in Engineering

Data Scientist and Developer

Location
Aix-les-Bains, France
Toptal Member Since
December 14, 2021

Xavier holds a PhD in particle physics and participates in one of the two main CERN collaboration projects. For five years now, he has been working as a postdoctoral researcher at Brown University (USA) and RWTH Aachen University (Germany), developing in-depth knowledge about deep learning. Xavier studied the performance of several generations of algorithms for 3D object identification and worked on medical imaging and genomics projects, eager to apply data analysis to medicine.

Portfolio

ICANS
Python 3, Kedro, Streamlit, Analysis, Genowe Wide Association Study, Statistics...
RWTH Aachen University
Management, Explainable Artificial Intelligence (XAI), Safety...
Brown University
Data Analysis, Deep Learning, Anomaly Detection, Python 3, Data Science...

Experience

Availability

Part-time

Preferred Environment

MacOS, Python, PyTorch, Django, Scikit-learn, Linux, Kedro

The most amazing...

...particle physics project I've worked on was to study the performance of several generations of deep learning algorithms for complex 3D object identification.

Work Experience

Biostatistician

2023 - PRESENT
ICANS
  • Performed genome-wide association studies related to breast cancer. Created a Kedro workflow allowing for easy re-use of the whole analysis pipeline and possible re-cast.
  • Reviewed from methodological and statistical perspectives over 30 clinical trial protocols at various stages of development, from idea to submission to regulatory bodies.
  • Trained medical and PhD students and postdocs in data analysis and statistics. Built streamlit application, allowing clinical trial analyses (survival, hypothesis testing, etc.) to be run efficiently.
Technologies: Python 3, Kedro, Streamlit, Analysis, Genowe Wide Association Study, Statistics, Teaching, Research

Postdoctoral Researcher

2017 - 2022
RWTH Aachen University
  • Managed a group of physicists in charge of 3D object identification within the CMS collaboration for two years. Defined the group's priorities and ensured the continuous development of new algorithms and calibration techniques.
  • Initiated a new calibration method to account for the difference between data taken at the Large Hadron Collider (LHC) and the simulation.
  • Contributed to creating a new physics analysis targeting the study of Higgs boson coupling to a specific particle.
  • Supervised bachelor, master, and PhD students while working on vertexing and object identification. Helped define scientific projects based on new developments in the field of deep learning.
Technologies: Management, Explainable Artificial Intelligence (XAI), Artificial Intelligence (AI), Safety, Deep Learning, Python 3, English, Statistics, Algorithms, Technical Leadership, Statistical Data Analysis, Python, Scikit-learn, Particle Physics, Physics, Data Analysis, Machine Learning, Data Visualization, Linux, Data Science, Big Data, Scientific Data Analysis, Anaconda, NumPy, Matplotlib, Pandas, MySQL, Research

Postdoctoral Researcher

2017 - 2022
Brown University
  • Initiated an effort to semi-automate data quality monitoring and data certification using dimension reduction within a detector group of the CMS collaboration.
  • Defined the structure of a new anomaly detection playground in the scope of a Django-based project to provide a platform for the comparison of various approaches to data certification automation.
  • Contributed to physics analyses targeting the study of the Higgs boson properties.
  • Supervised PhD students who were working on vertexing and object identification. Helped define scientific projects based on new developments in the field of deep learning.
Technologies: Data Analysis, Deep Learning, Anomaly Detection, Python 3, Data Science, English, Selenium, Statistics, Algorithms, Web Scraping, Technical Leadership, Statistical Data Analysis, Python, Scikit-learn, Particle Physics, Physics, Dimensionality Reduction, Machine Learning, Data Visualization, Management, Linux, Principal Component Analysis (PCA), Big Data, Scientific Data Analysis, Anaconda, NumPy, Matplotlib, Pandas, SQL, Artificial Intelligence (AI), MySQL, Research

Consultant

2021 - 2021
Freelance
  • Developed an algorithm for early anomaly detection in the medical imaging process.
  • Improved the potential patient management by allowing the imaging to stop early when an anomaly is detected.
  • Compared expert knowledge and machine learning approaches to anomaly detection.
Technologies: Python, Machine Learning, Dimensionality Reduction, Algorithms, Statistical Data Analysis, Python 3, Scikit-learn, Data Analysis, Data Visualization, Data Science, Anaconda, NumPy, Matplotlib, Pandas, Artificial Intelligence (AI)

PhD

2014 - 2017
Université de Strasbourg
  • Collaborated on a complex physics analysis, the study of the Higgs boson, adding a new analytical approach.
  • Contributed to the development of a complex object identification algorithm.
  • Led a group of physicists in charge of deploying object identification algorithms for data taking.
Technologies: C++, Python, Big Data, Scientific Computing, Scientific Data Analysis, Science, Communication, Statistics, Algorithms, Statistical Data Analysis, Python 3, Scikit-learn, Particle Physics, Physics, Data Analysis, Machine Learning, Data Visualization, Management, Linux, Data Science, Anaconda, NumPy, Matplotlib, Artificial Intelligence (AI), Research

Medical Imaging Anomaly Detection

Developed an algorithm for anomaly detection in medical imaging. Compared the performance between a knowledge-based approach and various machine learning approaches. Created a medical imaging dataset that could be useful for future studies.

Anomaly Detection Playground

https://github.com/CMSTrackerDPG/MLplayground
A Django and Kedro-based project that allows comparing various machine learning approaches to automate data certification. Provided a unique framework to compare the work of multiple physicists using PCA, NMF, and autoencoders for dimensionality reduction and anomaly detection.

Heavy Flavour Tagging

https://moriond.in2p3.fr/2019/EW/slides/2_Monday/2_afternoon/7_Coubez.pdf
An in-depth study of the performance of several generations of deep neural network-based algorithms used for complex 3D object identification. Developed new calibration techniques to correct differences between data and the simulation.

Languages

Python, Python 3, C++, SQL

Libraries/APIs

Pandas, NumPy, Matplotlib, Beautiful Soup, PyTorch, Scikit-learn

Paradigms

Data Science, Anomaly Detection, Management

Other

Particle Physics, Data Analysis, Machine Learning, Research, Deep Learning, English, APIs, Statistics, Algorithms, Technical Leadership, Statistical Data Analysis, Artificial Intelligence (AI), Science, Communication, Nuclear Physics, Physics, Medical Imaging, Generative Adversarial Networks (GANs), Dimensionality Reduction, Data Visualization, Explainable Artificial Intelligence (XAI), Outreach, Principal Component Analysis (PCA), Big Data, Scientific Computing, Scientific Data Analysis, Web Scraping, CI/CD Pipelines, Kedro, Safety, Analysis, Genowe Wide Association Study

Platforms

Anaconda, MacOS, Linux, Docker, Kubernetes

Frameworks

Selenium, Django, Streamlit

Storage

MySQL

Industry Expertise

Teaching

2014 - 2017

Ph.D. Degree in Particle Physics

University of Strasbourg - Strasbourg, France

2012 - 2014

Master's Degree in Subatomic Physics and Astroparticles

University of Strasbourg - Strasbourg, France

2009 - 2012

Bachelor's Degree in Physics

University of Strasbourg - Strasbourg, France

DECEMBER 2020 - PRESENT

Generative Adversarial Network (GANs) Specialization

DeepLearning.AI via Coursera

AUGUST 2020 - PRESENT

AI for Medicine Specialization

DeepLearning.AI via Coursera

AUGUST 2019 - PRESENT

Deep Learning Specialization

DeepLearning.AI via Coursera

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.

1

Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring