Matthias De Lozzo

Research Engineer
in Statistical Modelling & Artificial Intelligence

Who am I?

"Trained in data science, I have specialized in industrial statistics through applied research since 2010. With the diversity of my career in R&D, I want to put my skills as a data craftsman at the service of tomorrow's world. I am also open to an entrepreneurial adventure around data recovery." - MDL

Matthias DE LOZZO

31 years old, from Toulouse (France)
Currently in Toulouse, after two years of experience in Provence

My LinkedIn profile - My ResearchGate profile

Engineer - Ph. D in Statistical Modeling

Education: INSA Toulouse (engineer since 2010 - Ph. D since 2013)

Since november 2018: research engineer at IRT Saint Exupéry
October 2016 - October 2018: postdoctoral researcher at CERFACS (Centre Européen de Recherche et de Formation Avancée en Calcul Scientifique)
April 2014 - March 2016 : postdoctoral researcher at CEA (Commissariat à l'énergie atomique et aux énergies alternatives) de Cadarache

Qualification (MCF) - Section 26 (Applied mathematics and applications of mathematics)

Working experience: CNRM - Météo France, ONERA, Epsilon-Ingénierie, CEA and now CERFACS.

Interested in a wide range of statistical applications: environment, energy, economy, social, industry, health, text mining, big data, etc.

Professional interest for research and development studies ♦ multifaceted problems and applications ♦ applied and valued research ♦ project management and support ♦ training and transmission of knowledge ♦ discovering new fields of application for statistics ♦ innovation and human progress..

For more information, I invite you to read the rest of my resume online, or to download it in PDF format:

Curriculum Vitæ

Field of expertise


  • Statistical learning
  • Data Science
  • Machine learning
  • Data mining
  • Regression models
  • Decision models
  • Mixed models, lifetime models
  • Design of experiments and resampling
  • Statistical tests
  • ...

Research topics:

  • Uncertainty management
  • Sensitivity analysis
  • Surrogate models
  • Gaussian process models

Computer skills

Misc: LaTeX ♦ (Open) Office ♦ OpenTURNS ♦ RST ♦ R Shiny ♦ Slack

Programming: C ♦ C++ ♦ Fortran ♦ Git ♦ Java ♦ Matlab ♦R ♦ SAS ♦ Scilab ♦ SQL

OS: Linux ♦ Windows


Language skills

English: business fluent

Spanish: elementary

French: mother tongue


Vegetable cooking ♦ Environment ♦ Mountain ♦ Society ♦ Webmastering

Working experience

Since 11/2018

Research engineer at IRT Saint Exupéry, Toulouse, France (MDA-MDO team lead by Anne Gazaix)
Development of uncertainty quantification methodologies for problems of multidisciplinary design optimization and multidisciplinary analysis
  • Keywords: optimization, MDO, MDA, uncertainty quantification, surrogate models, numerical simulators, machine learning, GIT, Python.
10/2016 - 10/2018

Research engineer at CERFACS, Toulouse, France (GLOBC and ALGO teams), in collaboration with Sophie Ricci and Mélanie Rochoux
Development of uncertainty quantification methodologies for functional input and output parameters of computer codes
  • Keywords: uncertainty quantification, data assimilation, spatio-temporal random variables, numerical simulators, kriging, polynomial chaos expansion, machine learning, GIT, Python, R, SPOC, business softwares.
  • Applications: Management of uncertainties in the modelling of fluvial water heights subject to unsteady forcing.
  • Misc: Coaching of trainees and doctoral students ♦ Development of a SPOC in uncertainty quantification ♦ Development of an open source Python library dedicated to uncertainty quantification

Teaching, supervision and scientific mediation

  • Scientific mediation: European Researchers' Night 2017.
  • Supervision: trainees from INSA, ENM and ENSEEIHT, and doctoral students.
04/2014 - 03/2016

Research engineer (2 years) at CEA, Cadarache, France (DEN/DER/SESI/LEMS), in collaboration with Amandine Marrel
Development of statistical tools for the sensitivity analysis of calculation codes
Application to thermal-hydraulic safety studies and impact calculations in nuclear engineering.
  • Keywords: uncertainty quantification, sensitivity analysis, Sobol' indices screening, dependance measures, spatio-temporal random variables, kriging, numerical simulators, machine learning, R, Matlab, business softwares.
  • Application to nuclear engineering: Sodium (Total instantaneous plugging of a subassembly - Maximum sheath temperature over the whole core); Environmental impact studies (Radionuclide transport).
  • Misc: 6 publications; Contribution and correction of technical notes; Traineer ; Web interface for sensitivity analysis in R.

Teaching / Supervision


Ph. D (3 years) at ONERA, Toulouse, France (DTIM)
Spatio-temporal and multi-fidelity surrogate models
Application to aeronautical and thermal engineering

Teaching / Supervision

  • Project Hippocampe - Introduction to research in mathematics for high school students, Toulouse, France - 3 days.
  • Study week "Mathématiques et Entreprise" with the GDR Mathématiques-Entreprises, AMIES and IMT, Data assimilation for for road traffic, Toulouse, France.
  • Courses in mathematics (directed and practical work), INSA Toulouse, France, about 110 hours :
    • Logic and types of reasoning, sets - Licence 1
    • Algebra and analysis - Licence 2
    • Statistical inference and statistical tests - Licence 3
Internship (5 months) at ONERA, Toulouse, France (DTIM), supervised by Patricia Klotz
Adaptative designs of experiments for surrogate models
  • Keywords: surrogate models, model readjustment, variable reduction, artificial neural networks.
Internship (4 months) at Centre National de Recherches Météorologiques - Météo France, Toulouse, supervised by Aurélien Ribes
Statistical modeling of climatological data
  • MKeywords: time series, statistical tests for quantiles and expectation, extrema values.


Qualification (MCF) - Section 26 (Applied mathematics and applications of mathematics).
Ph. D in applied mathematics from Université de Toulouse - INSA Toulouse, France.
Training in fluid mechanics as part of my doctorate - INSA, Toulouse, France.
Mathematics and Modeling Engineer, specialty Methods and Statistical Modelss - INSA Toulouse, France.
Master 2 Research in Mathematics & Applications, specialty Statistics, with honours - INSA Toulouse, France.
Baccalauréat S SVT, option Mathematics, with honours, L’Isle-Jourdain, France.


International journals

Multilevel Monte Carlo covariance estimation for the computation of Sobol' indices, P. Mycek, M. De Lozzo, submitted to Journal on Uncertainty Quantification, under review.
Comparison of Polynomial Chaos and Gaussian Process surrogates for uncertainty quantification and correlation estimation of spatially distributed open-channel steady flows, P. T. Roy, N. El Moçayd, S. Ricci, J.-C. Jouhaud, N. Goutal, M. De Lozzo, M. Rochoux, Stochastic Environmental Research and Risk Assessment.
Sensitivity Analysis of Spatial and/or Temporal Phenomena, A. Marrel, N. Saint-Geours, M. De Lozzo, In: Ghanem R., Higdon D., Owhadi H. (eds) Handbook of Uncertainty Quantification, Springer.
Sensitivity analysis with dependence and variance-based measures for spatio-temporal numerical simulators, M. De Lozzo, A. Marrel, Stochastic Environmental Research & Risk Assessment, 31(6), p. 1437–1453.
Estimation of the derivative-based global sensitivity measures using a Gaussian process metamodel, M. De Lozzo, A. Marrel, Journal on Uncertainty Quantification, 4(1), p. 708–738.
New improvements in the use of dependence measures for sensitivity analysis and screening, M. De Lozzo, A. Marrel, Journal of Statistical Computation and Simulation, 86(15), p. 3038–3058.
Advanced surrogate model and sensitivity analysis methods for SFR accident assessment, A. Marrel, N. Marie, M. De Lozzo, Reliability Engineering & System Safety, 138, p. 232–241.
Multilayer perceptron for the learning of spatio-temporal dynamics - Application in thermal engineering, M. De Lozzo, P. Klotz, B. Laurent, Engineering Applications of Artificial Intelligence, 26(10), p. 2270–2286.

French journals

Substitution de modèle et approche multifidélité en expérimentation numérique, M. De Lozzo, Journal de la Société Française de Statistique, 156(3), p. 21–55.


Méthodes de Monte-Carlo multi-niveaux pour la quantification d'incertitudes et l'assimilation de données - Application à la modélisation fluviale, M. De Lozzo, P. Mycek, S. Ricci, M. Rochoux, P. T. Roy, N. Goutal, 49èmes Journées de Statistique, SFDS, Avignon, du 29 mai au 2 juin 2017.
Uncertainty quantification for the Gironde estuary hydrodynamics with Telemac 2D, V. Laborie, N. Goutal, S. Ricci, M. De Lozzo, P. Sergent, SymHydro 2017.
Estimation des mesures de sensibilité globale basées sur les dérivées via un métamodèle par processus gaussien, M. De Lozzo, A. Marrel, 47e Journées de Statistique, SFDS, Lille, du 1er au 5 juin 2015.
Uncertainty propagation for reliability analysis: application to sodium fast reactor safety studies, A. Marrel, N. Seiler-Marie, M. De Lozzo, UNCECOMP, Hersonissos (Grèce), du 25 au 27 mai 2015.
Learning of Spatio-temporal Dynamics in Thermal Engineering, M. De Lozzo, P. Klotz, B. Laurent, Engineering Applications of Neural Networks dans Communications in Computer and Information Science, 311, p. 213–222.

Oral communications

Séminaire de l'unité MIAT de l'INRA de Toulouse, le 15 septembre 2017 : "Échantillonnage Monte-Carlo multi-niveaux pour la quantification d'incertitudes et l'analyse de sensibilité"
49èmes Journées de Statistique, SFDS, Avignon, du 29 mai au 2 juin 2017 : "Méthodes de Monte-Carlo multi-niveaux pour la quantification d'incertitudes et l'assimilation de données - Application à la modélisation fluviale".
47e Journées de Statistique, SFDS, Lille, du 1er au 5 juin 2015 : "Estimation des mesures de sensibilité globale basées sur les dérivées via un métamodèle par processus gaussien"
Séminaire LJK-Statistique, Grenoble, le 18 décembre 2014 : "Mesures de dépendance pour l'analyse de sensibilité et le criblage : application aux simulateurs numériques"
45e Journées de Statistique, SFDS, Toulouse, du 27 au 31 mai 2013 : "Apprentissage de données multi-fidélités par mélange de processus gaussiens"
Groupe de travail du GdR MASCOT-NUM, Paris, le 17 mai 2013 : "Agrégation de co-krigeages pour les codes de calcul multi-fidélité"
Journées des Thèses ONERA 2013, Toulouse, du 31 janvier au 1er février 2013 : "Constructions de modèles de substitution multifidélité - Application à la simulation thermique"
13th International Conference, EANN 2012, Londres, du 20 au 23 septembre 2012 : "Learning of Spatio-temporal Dynamics in Thermal Engineering"
Journées du GdR MASCOT-NUM 2012, Bruyères-le-Châtel, du 21 au 23 mars 2012 : "Apprentissage de dynamiques spatio-temporelles par réseaux de neurones artificiels"
Journées des Thèses ONERA 2012, Toulouse, du 8 au 9 février 2012 : "Constructions de modèles de substitution multifidélité - Application à la simulation thermique"
Journées du GdR MASCOT-NUM 2011, Villard de Lans, du 23 au 25 mars 2011 : "Construction de modèles de substitution multi-fidélité - Application à la simulation thermique"
Journées des Thèses ONERA 2011, Toulouse, du 25 au 27 janvier 2011 : "Constructions de modèles de substitution multifidélité - Application à la simulation thermique"
Journée des doctorants en statistique de l'IMT 2010, Toulouse, le 14 septembre 2010 : "Plans d'expériences adaptatifs pour la construction de modèles de substitution"

Would you like to contact me?

Please leave a message via the form below
and I will answer you as soon as possible!