{"id":64685,"date":"2021-04-26T14:34:47","date_gmt":"2021-04-26T13:34:47","guid":{"rendered":"https:\/\/datascientest.com\/?p=64685"},"modified":"2023-11-09T14:07:35","modified_gmt":"2023-11-09T14:07:35","slug":"mlops-le-devops-applique-aux-projets-de-machine-learning","status":"publish","type":"post","link":"https:\/\/datascientest.com\/mlops-le-devops-applique-aux-projets-de-machine-learning","title":{"rendered":"MLOps : Le DevOps appliqu\u00e9 aux projets de Machine Learning"},"content":{"rendered":"\t\t<div data-elementor-type=\"wp-post\" data-elementor-id=\"64685\" class=\"elementor elementor-64685\" data-elementor-post-type=\"post\">\n\t\t\t\t\t\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-410d565 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"410d565\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-14a9985\" data-id=\"14a9985\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-5b378d8 elementor-widget elementor-widget-heading\" data-id=\"5b378d8\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<style>\/*! elementor - v3.15.0 - 02-08-2023 *\/\n.elementor-heading-title{padding:0;margin:0;line-height:1}.elementor-widget-heading .elementor-heading-title[class*=elementor-size-]>a{color:inherit;font-size:inherit;line-height:inherit}.elementor-widget-heading .elementor-heading-title.elementor-size-small{font-size:15px}.elementor-widget-heading .elementor-heading-title.elementor-size-medium{font-size:19px}.elementor-widget-heading .elementor-heading-title.elementor-size-large{font-size:29px}.elementor-widget-heading .elementor-heading-title.elementor-size-xl{font-size:39px}.elementor-widget-heading .elementor-heading-title.elementor-size-xxl{font-size:59px}<\/style><h2 class=\"elementor-heading-title elementor-size-default\">Dans un pr\u00e9c\u00e9dent article, nous pr\u00e9sentions la philosophie DevOps, et comment cette nouvelle approche permet une livraison de valeur plus rapide pour les entreprises au travers de l\u2019unification des \u00e9quipes de d\u00e9veloppement (Dev) et des op\u00e9rations (Ops), qui travaillaient pr\u00e9alablement en silos. Dans cet article, nous allons nous int\u00e9resser \u00e0 l\u2019application de cette approche dans le cadre de probl\u00e9matiques de Machine Learning: on parle de MLOps.<\/h2>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-2e5deab elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"2e5deab\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-8a4a1fd\" data-id=\"8a4a1fd\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-6d23679 elementor-widget elementor-widget-heading\" data-id=\"6d23679\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h3 class=\"elementor-heading-title elementor-size-default\">Retour sur le Devops<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-e20591c elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"e20591c\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-ebf1511\" data-id=\"ebf1511\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-b583605 elementor-widget elementor-widget-text-editor\" data-id=\"b583605\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<style>\/*! elementor - v3.15.0 - 02-08-2023 *\/\n.elementor-widget-text-editor.elementor-drop-cap-view-stacked .elementor-drop-cap{background-color:#69727d;color:#fff}.elementor-widget-text-editor.elementor-drop-cap-view-framed .elementor-drop-cap{color:#69727d;border:3px solid;background-color:transparent}.elementor-widget-text-editor:not(.elementor-drop-cap-view-default) .elementor-drop-cap{margin-top:8px}.elementor-widget-text-editor:not(.elementor-drop-cap-view-default) .elementor-drop-cap-letter{width:1em;height:1em}.elementor-widget-text-editor .elementor-drop-cap{float:left;text-align:center;line-height:1;font-size:50px}.elementor-widget-text-editor .elementor-drop-cap-letter{display:inline-block}<\/style>\t\t\t\t<p><span style=\"font-weight: 400;\">Le <strong><a href=\"https:\/\/datascientest.com\/devops\">DevOps<\/a><\/strong> permet une <\/span><b>r\u00e9duction du time to market<\/b><span style=\"font-weight: 400;\"> ainsi qu\u2019une am\u00e9lioration de la qualit\u00e9 des produits logiciels gr\u00e2ce \u00e0 <\/span><b>3 principes:<\/b><\/p><ul><li><b>D\u00e9silotage<\/b><span style=\"font-weight: 400;\"> : les \u00e9quipes de d\u00e9veloppement et les op\u00e9rations doivent travailler de concert, vers un objectif commun, en partageant toute l\u2019information.<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><b>Automatisation<\/b><span style=\"font-weight: 400;\"> : Tout ce qui est automatisable doit l\u2019\u00eatre : build, test, d\u00e9ploiement. <\/span><span style=\"font-weight: 400;\">Ceci dans un double objectif de r\u00e9duction du temps n\u00e9cessaire au d\u00e9ploiement et du nombre de non-qualit\u00e9s, les \u00e9tapes manuelles \u00e9tant propices aux erreurs humaines.<\/span><\/li><li><b>Monitoring<\/b><span style=\"font-weight: 400;\"> : r\u00e9colter et surveiller les m\u00e9triques importantes, \u00e0 la fois c\u00f4t\u00e9 m\u00e9tier (nombre d\u2019utilisateurs connect\u00e9s, nombre de commandes, etc.) et c\u00f4t\u00e9 op\u00e9rations (pourcentage d\u2019utilisation CPU \/ RAM des serveurs, nombre d\u2019erreurs, etc.)<\/span><\/li><\/ul>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-28e8075 elementor-widget elementor-widget-image\" data-id=\"28e8075\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<style>\/*! elementor - v3.15.0 - 02-08-2023 *\/\n.elementor-widget-image{text-align:center}.elementor-widget-image a{display:inline-block}.elementor-widget-image a img[src$=\".svg\"]{width:48px}.elementor-widget-image img{vertical-align:middle;display:inline-block}<\/style>\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"600\" height=\"324\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2021\/04\/Image-26-04-2021-a\u0300-14.03.jpg\" class=\"attachment-large size-large wp-image-64693\" alt=\"\" srcset=\"https:\/\/datascientest.com\/wp-content\/uploads\/2021\/04\/Image-26-04-2021-a\u0300-14.03.jpg 600w, https:\/\/datascientest.com\/wp-content\/uploads\/2021\/04\/Image-26-04-2021-a\u0300-14.03-300x162.jpg 300w\" sizes=\"(max-width: 600px) 100vw, 600px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-41efca2 elementor-align-center button-principale elementor-widget elementor-widget-button\" data-id=\"41efca2\" data-element_type=\"widget\" data-widget_type=\"button.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-button-wrapper\">\n\t\t\t<a class=\"elementor-button elementor-button-link elementor-size-sm elementor-animation-grow\" href=\"https:\/\/datascientest.com\/formation-ml-ops\">\n\t\t\t\t\t\t<span class=\"elementor-button-content-wrapper\">\n\t\t\t\t\t\t<span class=\"elementor-button-text\">D\u00e9couvrir la formation MLOps de DataScientest<\/span>\n\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-fd4c377 elementor-widget elementor-widget-text-editor\" data-id=\"fd4c377\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Suite \u00e0 l&rsquo;adoption massive du DevOps et \u00e0 l\u2019augmentation significative de l\u2019utilisation du Machine Learning et de l\u2019IA en entreprise, les concepts DevOps ont \u00e9t\u00e9 repris dans les projets data. Si les objectifs et principes restent globalement les m\u00eames, certaines sp\u00e9cificit\u00e9s propres aux machine learning ont n\u00e9cessit\u00e9 la cr\u00e9ation d\u2019une approche propre : le <\/span><b>MLOps<\/b><span style=\"font-weight: 400;\">.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-5faa892 elementor-widget elementor-widget-heading\" data-id=\"5faa892\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h3 class=\"elementor-heading-title elementor-size-default\">Qu\u2019est ce que le MLOps ?<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-f8459e4 elementor-widget elementor-widget-text-editor\" data-id=\"f8459e4\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Selon une <\/span><strong><a href=\"https:\/\/cloud.google.com\/solutions\/machine-learning\/mlops-continuous-delivery-and-automation-pipelines-in-machine-learning\">documentation de google<\/a><\/strong><span style=\"font-weight: 400;\"> :<\/span><\/p><p><b><i>MLOps <\/i><\/b><i><span style=\"font-weight: 400;\">est une culture et une pratique d&rsquo;ing\u00e9nierie ML qui vise \u00e0 unifier le d\u00e9veloppement (Dev) et les op\u00e9rations (Ops) des syst\u00e8mes de ML.<\/span><\/i><\/p><p><span style=\"font-weight: 400;\">Afin de comprendre ce qu\u2019est le MLOps, nous allons nous attacher \u00e0 saisir les diff\u00e9rences avec le DevOps classique :<\/span><\/p><ul><li><b>D\u00e9veloppement exp\u00e9rimental<\/b><span style=\"font-weight: 400;\"> : par nature, tous les projets ML suivent un d\u00e9veloppement non lin\u00e9aire : diff\u00e9rents <\/span><i><span style=\"font-weight: 400;\">preprocessing<\/span><\/i><span style=\"font-weight: 400;\">, <\/span><i><span style=\"font-weight: 400;\">feature engineering <\/span><\/i><span style=\"font-weight: 400;\">et algorithmes sont test\u00e9s, jusqu\u2019\u00e0 ce qu\u2019une performance suffisante soit atteinte. <\/span><span style=\"font-weight: 400;\">Garder une trace de ce qui a \u00e9t\u00e9 exp\u00e9riment\u00e9 et de ce qui a ou non fonctionn\u00e9 peut s\u2019av\u00e9rer complexe.<\/span><span style=\"font-weight: 400;\"> C\u2019est pourtant crucial pour r\u00e9duire le temps de d\u00e9veloppement, notamment dans le cas de grosses \u00e9quipes, afin d\u2019\u00e9viter de tester \u00e0 nouveau des approches qui auraient d\u00e9j\u00e0 \u00e9t\u00e9 exp\u00e9riment\u00e9es sans succ\u00e8s par un autre <\/span><i><span style=\"font-weight: 400;\">data scientist<\/span><\/i><span style=\"font-weight: 400;\">.<\/span><\/li><\/ul>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-489fdf3 elementor-widget elementor-widget-image\" data-id=\"489fdf3\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t<figure class=\"wp-caption\">\n\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"300\" height=\"298\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2021\/04\/Image-26-04-2021-a\u0300-14.58-300x298.jpg\" class=\"attachment-medium size-medium wp-image-64713\" alt=\"\u00e9tapes du d\u00e9veloppement d&#039;un mod\u00e8le de machine learning\" srcset=\"https:\/\/datascientest.com\/wp-content\/uploads\/2021\/04\/Image-26-04-2021-a\u0300-14.58-300x298.jpg 300w, https:\/\/datascientest.com\/wp-content\/uploads\/2021\/04\/Image-26-04-2021-a\u0300-14.58-1024x1017.jpg 1024w, https:\/\/datascientest.com\/wp-content\/uploads\/2021\/04\/Image-26-04-2021-a\u0300-14.58-150x150.jpg 150w, https:\/\/datascientest.com\/wp-content\/uploads\/2021\/04\/Image-26-04-2021-a\u0300-14.58-768x763.jpg 768w, https:\/\/datascientest.com\/wp-content\/uploads\/2021\/04\/Image-26-04-2021-a\u0300-14.58.jpg 1089w\" sizes=\"(max-width: 300px) 100vw, 300px\" \/>\t\t\t\t\t\t\t\t\t\t\t<figcaption class=\"widget-image-caption wp-caption-text\">\u00c9tapes du d\u00e9veloppement d'un mod\u00e8le de machine learning<\/figcaption>\n\t\t\t\t\t\t\t\t\t\t<\/figure>\n\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-8511cc4 elementor-widget elementor-widget-text-editor\" data-id=\"8511cc4\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<ul><li><b>D\u00e9rive conceptuelle<\/b><span style=\"font-weight: 400;\"> : \u00e0 la diff\u00e9rence d\u2019un produit logiciel classique, un syst\u00e8me de ML peut voir sa performance varier (et la plupart du temps d\u00e9cliner) au cours du temps, sans qu\u2019il y ait eu de modification dans le code ou l\u2019infrastructure. Ceci est d\u00fb au ph\u00e9nom\u00e8ne de <strong>d\u00e9rive conceptuelle<\/strong>, dans lequel les relations <strong>entre la variable \u00e0 pr\u00e9dire et les variables explicatives<\/strong> vont \u00e9voluer de mani\u00e8re impr\u00e9vue. <\/span><span style=\"font-weight: 400;\">Il est donc indispensable de suivre l\u2019\u00e9volution de la performance des mod\u00e8les de ML en production, et potentiellement d\u2019impl\u00e9menter une strat\u00e9gie de r\u00e9apprentissage automatique afin de \u00ab\u00a0re-calibrer\u00a0\u00bb le mod\u00e8le lorsque sa performance diminue.<\/span><span style=\"font-weight: 400;\"> On peut notamment citer l\u2019exemple de <strong>Google Flu Trends<\/strong>, un projet de Google qui devait permettre d\u2019estimer le nombre de cas de grippe \u00e0 partir des recherches effectu\u00e9es sur le moteur de recherche, mais qui <\/span><strong><a href=\"https:\/\/www.nature.com\/news\/when-google-got-flu-wrong-1.12413\">montra ses limites en surestimant largement le nombre de cas<\/a><\/strong><span style=\"font-weight: 400;\"> au bout de quelques ann\u00e9es d\u2019exploitations.<\/span><\/li><\/ul>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-ba13e44 elementor-widget elementor-widget-image\" data-id=\"ba13e44\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"196\" height=\"300\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2021\/04\/Image-26-04-2021-a\u0300-15.06-196x300.jpg\" class=\"attachment-medium size-medium wp-image-64715\" alt=\"D\u00e9monstration de Google Flu Trends\" srcset=\"https:\/\/datascientest.com\/wp-content\/uploads\/2021\/04\/Image-26-04-2021-a\u0300-15.06-196x300.jpg 196w, https:\/\/datascientest.com\/wp-content\/uploads\/2021\/04\/Image-26-04-2021-a\u0300-15.06.jpg 500w\" sizes=\"(max-width: 196px) 100vw, 196px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-52613f4 elementor-widget elementor-widget-text-editor\" data-id=\"52613f4\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<ul><li style=\"font-weight: 400;\" aria-level=\"1\"><b>\u00c9quipe<\/b><span style=\"font-weight: 400;\"> : Dans la plupart des organisations, les \u00e9quipes data transverses ou int\u00e9gr\u00e9es dans diff\u00e9rents d\u00e9partements m\u00e9tier sont relativement jeunes et manquent de ressources qualifi\u00e9es pour g\u00e9rer le d\u00e9ploiement et le maintien en condition op\u00e9rationnelle de <\/span><b>syst\u00e8mes ML complexes<\/b><span style=\"font-weight: 400;\">. En effet, ces \u00e9quipes se composent g\u00e9n\u00e9ralement principalement de data scientists qui se concentrent sur le d\u00e9veloppement des mod\u00e8les de machine learning, mais n\u2019ont pas les comp\u00e9tences n\u00e9cessaires pour g\u00e9rer le d\u00e9ploiement et la maintenance d\u2019applications compl\u00e8tes. <\/span><span style=\"font-weight: 400;\">De plus, les \u00e9quipes data \u00e9voluent encore trop souvent en silo, sans communiquer avec les diff\u00e9rentes \u00e9quipes techniques avec lesquelles elles devraient interagir pour mettre en production leurs mod\u00e8les.<\/span><\/li><\/ul><p><span style=\"font-weight: 400;\">Ces sp\u00e9cificit\u00e9s et complexit\u00e9s inh\u00e9rentes aux projets data expliquent en partie pourquoi, alors que de plus en plus d\u2019entreprises investissent massivement dans le machine learning et l\u2019IA, tr\u00e8s peu de mod\u00e8les sont effectivement d\u00e9ploy\u00e9s en production. <\/span><span style=\"font-weight: 400;\">Selon <\/span><strong><a href=\"https:\/\/info.algorithmia.com\/hubfs\/2019\/Whitepapers\/The-State-of-Enterprise-ML-2020\/Algorithmia_2020_State_of_Enterprise_ML.pdf\">une \u00e9tude<\/a><\/strong><span style=\"font-weight: 400;\"> men\u00e9e par Algorithmia en 2020, 55% des entreprises engag\u00e9es dans l\u2019IA n\u2019ont jamais r\u00e9alis\u00e9 de passage en production.<\/span><\/p><p><span style=\"font-weight: 400;\">Le MLOps a donc pour objectif de pallier ces difficult\u00e9s afin de permettre le d\u00e9ploiement et le <\/span><i><span style=\"font-weight: 400;\">run<\/span><\/i><span style=\"font-weight: 400;\"> des syst\u00e8mes ML, et ce en appliquant des principes similaires au DevOps :<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-02989af elementor-widget elementor-widget-text-editor\" data-id=\"02989af\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<ul><li aria-level=\"1\"><b>Monitoring<\/b><span style=\"font-weight: 400;\"> : Le monitoring prend encore de l\u2019importance dans la philosophie MLOps, en <\/span><i><span style=\"font-weight: 400;\">run <\/span><\/i><span style=\"font-weight: 400;\">(notamment en collectant les m\u00e9triques de performance des mod\u00e8les afin de d\u00e9tecter les d\u00e9rives conceptuelles), comme au stade de d\u00e9veloppement (pour traquer les diff\u00e9rentes exp\u00e9rimentations et leurs r\u00e9sultats).<\/span><\/li><\/ul><ul><li aria-level=\"1\"><b>Culture<\/b><span style=\"font-weight: 400;\"> : les \u00e9quipes data doivent \u00e9galement travailler avec les autres \u00e9quipes techniques (les op\u00e9rations mais \u00e9galement les \u00e9quipes d\u00e9veloppant les produits logiciels avec lesquels les mod\u00e8les de ML doivent s\u2019int\u00e9grer)<\/span><\/li><\/ul><ul><li aria-level=\"1\"><b>Automatisation<\/b><span style=\"font-weight: 400;\"> : En plus des \u00e9tapes classiques (build, test, deploy), le MLOps pr\u00f4ne l\u2019automatisation des t\u00e2ches propres \u00e0 la maintenance des syst\u00e8mes ML, comme le r\u00e9apprentissage des mod\u00e8les. Apr\u00e8s que les mod\u00e8les sont entra\u00een\u00e9s, ils doivent \u00e9galement \u00eatre automatiquement d\u00e9ploy\u00e9s.<\/span><\/li><\/ul><p><span style=\"font-weight: 400;\">Pour aider les \u00e9quipes data \u00e0 adopter une approche MLOps, de plus en plus d\u2019outils ont \u00e9t\u00e9 d\u00e9velopp\u00e9s. On pourra notamment citer <\/span><strong><a href=\"https:\/\/mlflow.org\/\">MLflow<\/a>, <a href=\"https:\/\/docs.metaflow.org\/introduction\/why-metaflow\">Metaflow<\/a>, <a href=\"https:\/\/www.kubeflow.org\/\">Kubeflow<\/a>.<\/strong><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-e443c1f elementor-widget__width-initial elementor-widget elementor-widget-video\" data-id=\"e443c1f\" data-element_type=\"widget\" data-settings=\"{&quot;youtube_url&quot;:&quot;https:\\\/\\\/youtu.be\\\/BHt3Z0IA8DE&quot;,&quot;video_type&quot;:&quot;youtube&quot;,&quot;controls&quot;:&quot;yes&quot;}\" data-widget_type=\"video.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<style>\/*! elementor - v3.15.0 - 02-08-2023 *\/\n.elementor-widget-video .elementor-widget-container{overflow:hidden;transform:translateZ(0)}.elementor-widget-video .elementor-wrapper{aspect-ratio:var(--video-aspect-ratio)}.elementor-widget-video .elementor-wrapper iframe,.elementor-widget-video .elementor-wrapper video{height:100%;width:100%;display:flex;border:none;background-color:#000}@supports not (aspect-ratio:1\/1){.elementor-widget-video .elementor-wrapper{position:relative;overflow:hidden;height:0;padding-bottom:calc(100% \/ var(--video-aspect-ratio))}.elementor-widget-video .elementor-wrapper iframe,.elementor-widget-video .elementor-wrapper video{position:absolute;top:0;right:0;bottom:0;left:0}}.elementor-widget-video .elementor-open-inline .elementor-custom-embed-image-overlay{position:absolute;top:0;right:0;bottom:0;left:0;background-size:cover;background-position:50%}.elementor-widget-video .elementor-custom-embed-image-overlay{cursor:pointer;text-align:center}.elementor-widget-video .elementor-custom-embed-image-overlay:hover .elementor-custom-embed-play i{opacity:1}.elementor-widget-video .elementor-custom-embed-image-overlay img{display:block;width:100%;aspect-ratio:var(--video-aspect-ratio);-o-object-fit:cover;object-fit:cover;-o-object-position:center center;object-position:center center}@supports not (aspect-ratio:1\/1){.elementor-widget-video .elementor-custom-embed-image-overlay{position:relative;overflow:hidden;height:0;padding-bottom:calc(100% \/ var(--video-aspect-ratio))}.elementor-widget-video .elementor-custom-embed-image-overlay img{position:absolute;top:0;right:0;bottom:0;left:0}}.elementor-widget-video .e-hosted-video .elementor-video{-o-object-fit:cover;object-fit:cover}.e-con-inner>.elementor-widget-video,.e-con>.elementor-widget-video{width:var(--container-widget-width);--flex-grow:var(--container-widget-flex-grow)}<\/style>\t\t<div class=\"elementor-wrapper elementor-open-inline\">\n\t\t\t<div class=\"elementor-video\"><\/div>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-5ca92ef elementor-widget elementor-widget-text-editor\" data-id=\"5ca92ef\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Mais pour pleinement r\u00e9ussir son virage MLOps il faudra avant tout casser les silos entre les op\u00e9rations et les \u00e9quipes data, et s\u2019assurer que celles-ci disposent de ressources n\u00e9cessaires (<\/span><strong><a href=\"https:\/\/datascientest.com\/formation-data-engineer\">Data Engineers<\/a>, <\/strong><a href=\"https:\/\/datascientest.com\/formation-machine-learning-engineer\"><span style=\"font-weight: 400;\"><strong>Machine Learning Engineers<\/strong><\/span><\/a><span style=\"font-weight: 400;\">). En effet, les Machine Learning Engineers sont des profils exp\u00e9riment\u00e9s qui disposent \u00e0 la fois d\u2019une expertise en machine learning (mod\u00e9lisation statistique, deep learning, etc.) mais aussi de comp\u00e9tences en d\u00e9veloppement logiciel, data engineering, et en passage en production des mod\u00e8les de ML.<\/span><\/p><p><span style=\"font-weight: 400;\">Vous souhaitez <\/span><b>devenir MLOps<\/b><span style=\"font-weight: 400;\"> ? N\u2019attendez plus, d\u00e9couvrez notre <a href=\"https:\/\/datascientest.com\/formation-ml-ops\">formation MLOps<\/a> !\u00a0<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-4489b6e elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"4489b6e\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-e7e8a37\" data-id=\"e7e8a37\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-745b602 elementor-align-center button-principale elementor-widget elementor-widget-button\" data-id=\"745b602\" data-element_type=\"widget\" data-widget_type=\"button.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-button-wrapper\">\n\t\t\t<a class=\"elementor-button elementor-button-link elementor-size-sm elementor-animation-grow\" href=\"https:\/\/datascientest.com\/formation-ml-ops\">\n\t\t\t\t\t\t<span class=\"elementor-button-content-wrapper\">\n\t\t\t\t\t\t<span class=\"elementor-button-text\">En savoir plus sur la formation MLOps<\/span>\n\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-70eda6b elementor-widget elementor-widget-html\" data-id=\"70eda6b\" data-element_type=\"widget\" data-widget_type=\"html.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<script type=\"application\/ld+json\">\n{\n  \"@context\": \"https:\/\/schema.org\",\n  \"@type\": \"FAQPage\",\n  \"mainEntity\": [{\n    \"@type\": \"Question\",\n    \"name\": \"Retour sur le Devops\",\n    \"acceptedAnswer\": {\n      \"@type\": \"Answer\",\n      \"text\": \"Le DevOps permet une r\u00e9duction du time to market ainsi qu\u2019une am\u00e9lioration de la qualit\u00e9 des produits logiciels gr\u00e2ce \u00e0 3 principes: D\u00e9silotage : les \u00e9quipes de d\u00e9veloppement et les op\u00e9rations doivent travailler de concert, vers un objectif commun, en partageant toute l\u2019information. Automatisation : Tout ce qui est automatisable doit l\u2019\u00eatre : build, test, d\u00e9ploiement. Ceci dans un double objectif de r\u00e9duction du temps n\u00e9cessaire au d\u00e9ploiement et du nombre de non-qualit\u00e9s, les \u00e9tapes manuelles \u00e9tant propices aux erreurs humaines. Monitoring : r\u00e9colter et surveiller les m\u00e9triques importantes, \u00e0 la fois c\u00f4t\u00e9 m\u00e9tier (nombre d\u2019utilisateurs connect\u00e9s, nombre de commandes, etc.) et c\u00f4t\u00e9 op\u00e9rations (pourcentage d\u2019utilisation CPU \/ RAM des serveurs, nombre d\u2019erreurs, etc.)\"\n    }\n  },{\n    \"@type\": \"Question\",\n    \"name\": \"Qu\u2019est ce que le MLOps ?\",\n    \"acceptedAnswer\": {\n      \"@type\": \"Answer\",\n      \"text\": \"elon une documentation de google : MLOps est une culture et une pratique d'ing\u00e9nierie ML qui vise \u00e0 unifier le d\u00e9veloppement (Dev) et les op\u00e9rations (Ops) des syst\u00e8mes de ML.\"\n    }\n  }]\n}\n<\/script>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>Dans un pr\u00e9c\u00e9dent article, nous pr\u00e9sentions la philosophie DevOps, et comment cette nouvelle approche permet une livraison de valeur plus rapide pour les entreprises au travers de l\u2019unification des \u00e9quipes de d\u00e9veloppement (Dev) et des op\u00e9rations (Ops), qui travaillaient pr\u00e9alablement en silos. Dans cet article, nous allons nous int\u00e9resser \u00e0 l\u2019application de cette approche dans [&hellip;]<\/p>\n","protected":false},"author":77,"featured_media":69574,"comment_status":"open","ping_status":"open","sticky":false,"template":"elementor_theme","format":"standard","meta":{"content-type":"","inline_featured_image":false},"categories":[2497],"tags":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO Premium plugin v21.7 (Yoast SEO v21.7) - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>MLOps : Le DevOps appliqu\u00e9 aux projets de Machine Learning<\/title>\n<meta name=\"description\" content=\"Dans cet article, nous allons nous int\u00e9resser \u00e0 l\u2019application de cette approche dans le cadre de probl\u00e9matiques de Machine Learning: on parle de MLOps.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/datascientest.com\/mlops-le-devops-applique-aux-projets-de-machine-learning\" \/>\n<meta property=\"og:locale\" content=\"fr_FR\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"MLOps : Le DevOps appliqu\u00e9 aux projets de Machine Learning\" \/>\n<meta property=\"og:url\" content=\"https:\/\/datascientest.com\/mlops-le-devops-applique-aux-projets-de-machine-learning\" \/>\n<meta property=\"og:site_name\" content=\"Formation Data Science | DataScientest.com\" \/>\n<meta property=\"article:published_time\" content=\"2021-04-26T13:34:47+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2023-11-09T14:07:35+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/datascientest.com\/wp-content\/uploads\/2021\/04\/illu_mlops_blog_2-02.png\" \/>\n\t<meta property=\"og:image:width\" content=\"1024\" \/>\n\t<meta property=\"og:image:height\" content=\"562\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"Raphael Kassel\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u00c9crit par\" \/>\n\t<meta name=\"twitter:data1\" content=\"Raphael Kassel\" \/>\n\t<meta name=\"twitter:label2\" content=\"Dur\u00e9e de lecture estim\u00e9e\" \/>\n\t<meta name=\"twitter:data2\" content=\"6 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/datascientest.com\/mlops-le-devops-applique-aux-projets-de-machine-learning\",\"url\":\"https:\/\/datascientest.com\/mlops-le-devops-applique-aux-projets-de-machine-learning\",\"name\":\"MLOps : Le DevOps appliqu\u00e9 aux projets de Machine Learning\",\"isPartOf\":{\"@id\":\"https:\/\/datascientest.com\/#website\"},\"datePublished\":\"2021-04-26T13:34:47+00:00\",\"dateModified\":\"2023-11-09T14:07:35+00:00\",\"author\":{\"@id\":\"https:\/\/datascientest.com\/#\/schema\/person\/6eca16f63daaafcbc96e15549b9d4ee1\"},\"description\":\"Dans cet article, nous allons nous int\u00e9resser \u00e0 l\u2019application de cette approche dans le cadre de probl\u00e9matiques de Machine Learning: on parle de MLOps.\",\"inLanguage\":\"fr-FR\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/datascientest.com\/mlops-le-devops-applique-aux-projets-de-machine-learning\"]}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/datascientest.com\/#website\",\"url\":\"https:\/\/datascientest.com\/\",\"name\":\"Formation Data Science | DataScientest.com\",\"description\":\"Leader de la formation en Data Science\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/datascientest.com\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"fr-FR\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/datascientest.com\/#\/schema\/person\/6eca16f63daaafcbc96e15549b9d4ee1\",\"name\":\"Raphael Kassel\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\/\/datascientest.com\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/raph.jpeg\",\"contentUrl\":\"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/raph.jpeg\",\"caption\":\"Raphael Kassel\"},\"description\":\"Raphael Kassel est le Head of Data Programs chez DataScientest. Il supervise une \u00e9quipe Data, impl\u00e9mente de nouveaux parcours comme \\\"Chef de projet IA\\\" et \\\"Data Product Manager\\\", et contribue \u00e0 la modernisation des op\u00e9rations du groupe. En collaboration \u00e9troite avec l'\u00e9quipe de d\u00e9veloppement, il adapte l'accompagnement acad\u00e9mique aux \u00e9volutions de l'\u00e9quipe data et \u00e0 l'internationalisation.\",\"sameAs\":[\"https:\/\/www.linkedin.com\/in\/raphaelkassel\/\",\"Data Science\"],\"url\":\"https:\/\/datascientest.com\/author\/raphaelkassel\"}]}<\/script>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"MLOps : Le DevOps appliqu\u00e9 aux projets de Machine Learning","description":"Dans cet article, nous allons nous int\u00e9resser \u00e0 l\u2019application de cette approche dans le cadre de probl\u00e9matiques de Machine Learning: on parle de MLOps.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/datascientest.com\/mlops-le-devops-applique-aux-projets-de-machine-learning","og_locale":"fr_FR","og_type":"article","og_title":"MLOps : Le DevOps appliqu\u00e9 aux projets de Machine Learning","og_url":"https:\/\/datascientest.com\/mlops-le-devops-applique-aux-projets-de-machine-learning","og_site_name":"Formation Data Science | DataScientest.com","article_published_time":"2021-04-26T13:34:47+00:00","article_modified_time":"2023-11-09T14:07:35+00:00","og_image":[{"width":1024,"height":562,"url":"https:\/\/datascientest.com\/wp-content\/uploads\/2021\/04\/illu_mlops_blog_2-02.png","type":"image\/png"}],"author":"Raphael Kassel","twitter_card":"summary_large_image","twitter_misc":{"\u00c9crit par":"Raphael Kassel","Dur\u00e9e de lecture estim\u00e9e":"6 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/datascientest.com\/mlops-le-devops-applique-aux-projets-de-machine-learning","url":"https:\/\/datascientest.com\/mlops-le-devops-applique-aux-projets-de-machine-learning","name":"MLOps : Le DevOps appliqu\u00e9 aux projets de Machine Learning","isPartOf":{"@id":"https:\/\/datascientest.com\/#website"},"datePublished":"2021-04-26T13:34:47+00:00","dateModified":"2023-11-09T14:07:35+00:00","author":{"@id":"https:\/\/datascientest.com\/#\/schema\/person\/6eca16f63daaafcbc96e15549b9d4ee1"},"description":"Dans cet article, nous allons nous int\u00e9resser \u00e0 l\u2019application de cette approche dans le cadre de probl\u00e9matiques de Machine Learning: on parle de MLOps.","inLanguage":"fr-FR","potentialAction":[{"@type":"ReadAction","target":["https:\/\/datascientest.com\/mlops-le-devops-applique-aux-projets-de-machine-learning"]}]},{"@type":"WebSite","@id":"https:\/\/datascientest.com\/#website","url":"https:\/\/datascientest.com\/","name":"Formation Data Science | DataScientest.com","description":"Leader de la formation en Data Science","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/datascientest.com\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"fr-FR"},{"@type":"Person","@id":"https:\/\/datascientest.com\/#\/schema\/person\/6eca16f63daaafcbc96e15549b9d4ee1","name":"Raphael Kassel","image":{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/datascientest.com\/#\/schema\/person\/image\/","url":"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/raph.jpeg","contentUrl":"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/raph.jpeg","caption":"Raphael Kassel"},"description":"Raphael Kassel est le Head of Data Programs chez DataScientest. Il supervise une \u00e9quipe Data, impl\u00e9mente de nouveaux parcours comme \"Chef de projet IA\" et \"Data Product Manager\", et contribue \u00e0 la modernisation des op\u00e9rations du groupe. En collaboration \u00e9troite avec l'\u00e9quipe de d\u00e9veloppement, il adapte l'accompagnement acad\u00e9mique aux \u00e9volutions de l'\u00e9quipe data et \u00e0 l'internationalisation.","sameAs":["https:\/\/www.linkedin.com\/in\/raphaelkassel\/","Data Science"],"url":"https:\/\/datascientest.com\/author\/raphaelkassel"}]}},"_links":{"self":[{"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/posts\/64685"}],"collection":[{"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/users\/77"}],"replies":[{"embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/comments?post=64685"}],"version-history":[{"count":0,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/posts\/64685\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/media\/69574"}],"wp:attachment":[{"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/media?parent=64685"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/categories?post=64685"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/tags?post=64685"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}