Experience

Gaiha

• France
• Biotechnology Research
• 1 - 100 Employee

• Chief Technology Officer (CTO)

  • Mar 2020 - Present

  Health project whose mission is to enable exceptional pre diagnosis to save every life and transform global healthcare. We look forward to supporting the medical patients' prioritization for the world's deadliest diseases and supporting healthcare professionals to focus on urgent cases. Gaiha is a medtech company specialized in artificial intelligence (AI) softwares for pre-diagnosis of several diseases: www.gaiha.org. Gaiha offers AI SaaS solutions, which are scientifically and… Show more Health project whose mission is to enable exceptional pre diagnosis to save every life and transform global healthcare. We look forward to supporting the medical patients' prioritization for the world's deadliest diseases and supporting healthcare professionals to focus on urgent cases. Gaiha is a medtech company specialized in artificial intelligence (AI) softwares for pre-diagnosis of several diseases: www.gaiha.org. Gaiha offers AI SaaS solutions, which are scientifically and clinically validated cutting edge technologies, for: - diabetic retinopathy detection, and its level of severity, and maculopathy detection based on fundus images - glaucoma detection - early breast cancer detection (i.e. frequent ER+ breast cancer type) based on blood sample and anthropometric measurements Show less Health project whose mission is to enable exceptional pre diagnosis to save every life and transform global healthcare. We look forward to supporting the medical patients' prioritization for the world's deadliest diseases and supporting healthcare professionals to focus on urgent cases. Gaiha is a medtech company specialized in artificial intelligence (AI) softwares for pre-diagnosis of several diseases: www.gaiha.org. Gaiha offers AI SaaS solutions, which are scientifically and… Show more Health project whose mission is to enable exceptional pre diagnosis to save every life and transform global healthcare. We look forward to supporting the medical patients' prioritization for the world's deadliest diseases and supporting healthcare professionals to focus on urgent cases. Gaiha is a medtech company specialized in artificial intelligence (AI) softwares for pre-diagnosis of several diseases: www.gaiha.org. Gaiha offers AI SaaS solutions, which are scientifically and clinically validated cutting edge technologies, for: - diabetic retinopathy detection, and its level of severity, and maculopathy detection based on fundus images - glaucoma detection - early breast cancer detection (i.e. frequent ER+ breast cancer type) based on blood sample and anthropometric measurements Show less



CYM - Predictive Maintenance

• France
• Software Development
• 1 - 100 Employee

• Chief Technology Officer (CTO)

  • Dec 2017 - Present

  What if tomorrow your systems would never break again? We support our customers in achieving better industrial performance thanks to AI applied to Maintenance. CYM’s SaaS solution is based on proprietary mathematical models, which enables to analyze clients’ large datasets to provide them: - Recommendation to perform early sign of drift and predictive maintenance (before failure occurrence). - Real-time manufacturing intelligence to monitor and optimize production line… Show more What if tomorrow your systems would never break again? We support our customers in achieving better industrial performance thanks to AI applied to Maintenance. CYM’s SaaS solution is based on proprietary mathematical models, which enables to analyze clients’ large datasets to provide them: - Recommendation to perform early sign of drift and predictive maintenance (before failure occurrence). - Real-time manufacturing intelligence to monitor and optimize production line performance (e.g. detection of over-consumption or under usage, etc.). Twitter @Cym_IoT Tél 01-86-86-15-15 CYM L'ElectroLab Hacker Space - 52, rue Paul Lescop 92000 NANTERRE Show less What if tomorrow your systems would never break again? We support our customers in achieving better industrial performance thanks to AI applied to Maintenance. CYM’s SaaS solution is based on proprietary mathematical models, which enables to analyze clients’ large datasets to provide them: - Recommendation to perform early sign of drift and predictive maintenance (before failure occurrence). - Real-time manufacturing intelligence to monitor and optimize production line… Show more What if tomorrow your systems would never break again? We support our customers in achieving better industrial performance thanks to AI applied to Maintenance. CYM’s SaaS solution is based on proprietary mathematical models, which enables to analyze clients’ large datasets to provide them: - Recommendation to perform early sign of drift and predictive maintenance (before failure occurrence). - Real-time manufacturing intelligence to monitor and optimize production line performance (e.g. detection of over-consumption or under usage, etc.). Twitter @Cym_IoT Tél 01-86-86-15-15 CYM L'ElectroLab Hacker Space - 52, rue Paul Lescop 92000 NANTERRE Show less



OKP4 - Open Knowledge Protocol for...

• France
• IT Services and IT Consulting
• 1 - 100 Employee

• Chief Data Officer (CDO)

  • Sep 2021 - Oct 2022

  OKP4 is a company building a decentralized protocol within the web3 framework, based on the COSMOS blockchain technology, which allows data sharing and retribution of new knowledge creation. I lead an amazing team of data scientists at this company where we develop many data, and deep-learning based services, which allow data sharing and knowledge creation. For example, we developed the following services: - A metadata extraction service for any file type; video, audio, images… Show more OKP4 is a company building a decentralized protocol within the web3 framework, based on the COSMOS blockchain technology, which allows data sharing and retribution of new knowledge creation. I lead an amazing team of data scientists at this company where we develop many data, and deep-learning based services, which allow data sharing and knowledge creation. For example, we developed the following services: - A metadata extraction service for any file type; video, audio, images, text, csv, etc. - A deep-learning based personal data detection service, which uses semantic analysis, for RGPD compliancy - A deep-learning translation service which allows translation between 18 languages - Etc. Show less OKP4 is a company building a decentralized protocol within the web3 framework, based on the COSMOS blockchain technology, which allows data sharing and retribution of new knowledge creation. I lead an amazing team of data scientists at this company where we develop many data, and deep-learning based services, which allow data sharing and knowledge creation. For example, we developed the following services: - A metadata extraction service for any file type; video, audio, images… Show more OKP4 is a company building a decentralized protocol within the web3 framework, based on the COSMOS blockchain technology, which allows data sharing and retribution of new knowledge creation. I lead an amazing team of data scientists at this company where we develop many data, and deep-learning based services, which allow data sharing and knowledge creation. For example, we developed the following services: - A metadata extraction service for any file type; video, audio, images, text, csv, etc. - A deep-learning based personal data detection service, which uses semantic analysis, for RGPD compliancy - A deep-learning translation service which allows translation between 18 languages - Etc. Show less



Capgemini

• India
• Information Services
• 1 - 100 Employee

• Data Science Researcher & Consultant

  • Jul 2016 - Nov 2017

  Development of several new algorithms and applications, or applying current state of the art methods, for industrial predictive maintenance and manufacturing intelligence. Development of several new algorithms and applications, or applying current state of the art methods, for industrial predictive maintenance and manufacturing intelligence.



CIRAD

• France
• Research Services
• 700 & Above Employee

• Research scientist (statistics and machine learning)

  • Nov 2014 - Dec 2016

  • Development of the CRAN library KRMM for solving "Reproducing Kernel Hilbert Space" (RKHS) regression, used by teams of INRA and CIRAD for genomic prediction and analyses, which competes with current implementations of SVM, random forests and neural networks. This package allows the use of several kernels ; linear, Gaussian, Laplacian, polynomial and ANOVA, when solving kernel ridge regression within the mixed model framework. • R package development, for simulating omic data… Show more • Development of the CRAN library KRMM for solving "Reproducing Kernel Hilbert Space" (RKHS) regression, used by teams of INRA and CIRAD for genomic prediction and analyses, which competes with current implementations of SVM, random forests and neural networks. This package allows the use of several kernels ; linear, Gaussian, Laplacian, polynomial and ANOVA, when solving kernel ridge regression within the mixed model framework. • R package development, for simulating omic data, based on LOESS regression and maximum likelihood optimized using the BFGS descent direction algorithm • Lecturer and organizer for an international genomic selection workshop : regularized linear regressions (ridge regression, LASSO, elastic-net), equivalences between Bayesian linear regressions and regularized linear models, RKHS and kernel methods (SVM regression, kernel ridge regression, Nadaraya-Watson) Show less • Development of the CRAN library KRMM for solving "Reproducing Kernel Hilbert Space" (RKHS) regression, used by teams of INRA and CIRAD for genomic prediction and analyses, which competes with current implementations of SVM, random forests and neural networks. This package allows the use of several kernels ; linear, Gaussian, Laplacian, polynomial and ANOVA, when solving kernel ridge regression within the mixed model framework. • R package development, for simulating omic data… Show more • Development of the CRAN library KRMM for solving "Reproducing Kernel Hilbert Space" (RKHS) regression, used by teams of INRA and CIRAD for genomic prediction and analyses, which competes with current implementations of SVM, random forests and neural networks. This package allows the use of several kernels ; linear, Gaussian, Laplacian, polynomial and ANOVA, when solving kernel ridge regression within the mixed model framework. • R package development, for simulating omic data, based on LOESS regression and maximum likelihood optimized using the BFGS descent direction algorithm • Lecturer and organizer for an international genomic selection workshop : regularized linear regressions (ridge regression, LASSO, elastic-net), equivalences between Bayesian linear regressions and regularized linear models, RKHS and kernel methods (SVM regression, kernel ridge regression, Nadaraya-Watson) Show less



INRA

• Germany
• Individual and Family Services

• PhD (statistics)

  • Oct 2011 - Oct 2014

  • Theoretical development and implementation (Fortran 90) of the entrywise 1-norm, associated to combined omic data (i.e. haplotype), for discriminating and optimizing predictive/clustering methods used for mapping of quantitative trait loci (QTL) • Implementation of a general EM algorithm (Fortran 90) for parametric random effects models • Development of shell and R scripts for parametric random effects models, and the associated restricted likelihood ratio test… Show more • Theoretical development and implementation (Fortran 90) of the entrywise 1-norm, associated to combined omic data (i.e. haplotype), for discriminating and optimizing predictive/clustering methods used for mapping of quantitative trait loci (QTL) • Implementation of a general EM algorithm (Fortran 90) for parametric random effects models • Development of shell and R scripts for parametric random effects models, and the associated restricted likelihood ratio test, for mapping of QTL • Implementation of different data generating processes in R (i.e. genetic models) for phenotype response simulation • Implementation of Monte-Carlo procedures for estimating statistical powers in R, associated to the different genetic models, and comparison with (ideal) theoretical statistical powers derived algebraically Show less • Theoretical development and implementation (Fortran 90) of the entrywise 1-norm, associated to combined omic data (i.e. haplotype), for discriminating and optimizing predictive/clustering methods used for mapping of quantitative trait loci (QTL) • Implementation of a general EM algorithm (Fortran 90) for parametric random effects models • Development of shell and R scripts for parametric random effects models, and the associated restricted likelihood ratio test… Show more • Theoretical development and implementation (Fortran 90) of the entrywise 1-norm, associated to combined omic data (i.e. haplotype), for discriminating and optimizing predictive/clustering methods used for mapping of quantitative trait loci (QTL) • Implementation of a general EM algorithm (Fortran 90) for parametric random effects models • Development of shell and R scripts for parametric random effects models, and the associated restricted likelihood ratio test, for mapping of QTL • Implementation of different data generating processes in R (i.e. genetic models) for phenotype response simulation • Implementation of Monte-Carlo procedures for estimating statistical powers in R, associated to the different genetic models, and comparison with (ideal) theoretical statistical powers derived algebraically Show less