CINXE.COM

<!DOCTYPE html>   <html class="no-js" lang="en-US">  <head>    <script type="text/javascript" src="https://www.databricks.com/sites/default/files/onetrust/DB_OtAutoBlock.js?v=1" ></script> <script src="https://cdn.cookielaw.org/scripttemplates/otSDKStub.js" data-document-language="true" type="text/javascript" charset="UTF-8" data-domain-script="92466579-1717-44d3-809d-a05fb02843ed" ></script> <script type="text/javascript" src="https://www.databricks.com/wp-content/plugins/databricks/js/onetrust.js?ver=1.0.0" id="db-onetrust-script" ></script> <link rel="stylesheet" id="db-onetrust-style" href="https://www.databricks.com/wp-content/uploads/db_onetrust.css" media="all" />   <script> window.rudderstackKey = "2SOR9fvSr5Fi6tN2ihPbVHnX1SZ"; </script>  <script type="text/javascript" src="https://www.databricks.com/sites/default/files/rudderstack/v1/db-rudderstack-events.js" ></script>  <meta charset="utf-8"> <meta http-equiv="X-UA-Compatible" content="IE=edge"> <meta http-equiv="X-UA-Compatible" content="IE=9" /> <meta content="Learn how Databricks uses MLflow to manage the end-to-end machine learning lifecycle." name="description" /> <meta name="viewport" content="width=device-width, initial-scale=1.0, maximum-scale=1.0, user-scalable=no" /> <meta name="viewport" content="width=device-width, initial-scale=1, maximum-scale=1, user-scalable=0"> <meta property="og:image" content="https://www.databricks.com/wp-content/uploads/2020/04/og-databricks.png"> <meta property="og:image:type" content="image/png"> <meta property="og:title" content="MLflow for gen AI agent and ML model lifecycle"> <meta property="og:image:width" content="1200"> <meta property="og:image:height" content="630"> <meta property="og:type" content="website"> <meta property="og:url" content="https://docs.databricks.com"> <meta property="og:description" content="" id="og-description"> <meta name="twitter:image" content="https://www.databricks.com/wp-content/uploads/2020/04/og-databricks.png"> <meta name="twitter:site" content="@databricks"> <meta name="twitter:creator" content="@databricks"> <meta property="twitter:description" content=""> <title>MLflow for gen AI agent and ML model lifecycle | Databricks on AWS</title> <link rel="canonical" href="https://docs.databricks.com/en/mlflow/index.html">  <link rel="alternate" hreflang="en" href="https://docs.databricks.com/en/mlflow/index.html" /> <link rel="alternate" hreflang="pt" href="https://docs.databricks.com/pt/mlflow/index.html" /> <link rel="alternate" hreflang="ja" href="https://docs.databricks.com/ja/mlflow/index.html" /> <link rel="alternate" hreflang="x-default" href="https://docs.databricks.com/en/mlflow/index.html" />  <link rel="shortcut icon" href="../_static/favicon.ico" />  <script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0], j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src= 'https://www.googletagmanager.com/gtm.js?id='+i+dl; j.setAttributeNode(d.createAttribute('data-ot-ignore')); f.parentNode.insertBefore(j,f); })(window,document,'script','dataLayer','GTM-T85FQ33');</script> <script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0], j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src= 'https://www.googletagmanager.com/gtm.js?id='+i+dl; j.setAttributeNode(d.createAttribute('data-ot-ignore')); f.parentNode.insertBefore(j,f); })(window,document,'script','dataLayer','GTM-TWTKQQ');</script>   <link rel="preconnect" href="https://fonts.googleapis.com"> <link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>  <link href="https://fonts.googleapis.com/css2?family=DM+Mono:ital@0;1&family=DM+Sans:ital,wght@0,400;0,700;1,400;1,700&display=swap" rel="stylesheet">  <link rel="stylesheet" type="text/css" href="../_static/css/theme.css?v=c44120be" /> <link rel="stylesheet" type="text/css" href="../_static/css/custom.css?v=21342a0b" /> <link rel="stylesheet" type="text/css" href="../_static/css/dropdown.css?v=46590757" /> <link rel="stylesheet" type="text/css" href="../_static/css/searchunify/main.css?v=96405b38" /> <script type="text/javascript"> // Set the static root for the theme window.STATIC_ROOT = `../_static`; var DOCUMENTATION_OPTIONS = { URL_ROOT: '../', VERSION: '1.0', COLLAPSE_INDEX: false, FILE_SUFFIX: '.html', HAS_SOURCE: 'false' }; </script>  <script defer data-ot-ignore type="text/javascript" src="../_static/jquery.js?v=5be48651"></script> <script defer data-ot-ignore type="text/javascript" src="../_static/underscore.js?v=ff3b9a9a"></script> <script defer data-ot-ignore type="text/javascript" src="../_static/doctools.js?v=695de88e"></script> <script defer data-ot-ignore type="text/javascript" src="../_static/language_data.js?v=a1340169"></script> <script defer data-ot-ignore type="text/javascript" src="../_static/js/vendor/jquery.waypoints.min.js?v=aa86c767"></script> <script defer data-ot-ignore type="text/javascript" src="../_static/js/localized.js?v=e71e2bb8"></script> <script defer data-ot-ignore type="text/javascript" src="../_static/js/clipboard.js?v=ba7e99af"></script> <script defer data-ot-ignore type="text/javascript" src="../_static/js/theme-nav.js?v=3af59ba7"></script> <script defer data-ot-ignore type="text/javascript" src="../_static/js/code-language-tabs.js?v=cb6ea2f5"></script> <script defer data-ot-ignore type="text/javascript" src="../_static/js/embedded-notebook.js?v=45f58fb2"></script> <script defer data-ot-ignore type="text/javascript" src="../_static/js/right-nav.js?v=301278ed"></script> <script defer data-ot-ignore type="text/javascript" src="../_static/js/custom.js?v=db7f62de"></script> <script defer data-ot-ignore type="text/javascript" src="../_static/js/searchunify/search-loader.js?v=05d3e7ac"></script> <script type="text/javascript"> // Replace this with sphinxext-opengraph window.onload = function () { var description = document .querySelector('meta[name="description"]') .getAttribute("content"); let titleText = document.querySelector("h1").textContent; document .querySelector('meta[property="og:title"]') .setAttribute("content", titleText); document .querySelector('meta[property="og:description"]') .setAttribute("content", description); document .querySelector('meta[property="twitter:description"]') .setAttribute("content", description); }; </script> <link rel="index" title="Index" href="../genindex.html" /> <link rel="search" title="Search" href="../search.html" /> <link rel="top" title="Databricks on AWS" href="../index.html" /> <link rel="up" title="AI and machine learning on Databricks" href="../machine-learning/index.html" /> <link rel="next" title="Track model development using MLflow" href="tracking.html" /> <link rel="prev" title="LLMOps workflows on Databricks" href="../machine-learning/mlops/llmops.html" /> </head> <body class="wy-body-for-nav" role="document">  <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-T85FQ33" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-TWTKQQ" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript>  <nav class="wy-nav-top header su_header" role="navigation" aria-label="top navigation"> <nav class="wy-nav-top header su_header" role="navigation" aria-label="top navigation"> <div class="container-logo"> <ul class="mobile-menu-toggle"> <li class="menu-toggle"> <i data-toggle="wy-nav-top" class="wy-nav-top-menu-button db-icon db-icon-menu pull-left"></i> <a href="https://www.databricks.com/" class="wy-nav-top-logo"><img src="../_static/small-scale-lockup-full-color-rgb.svg" width="137" height="21" alt="Databricks" /></a> </li> </ul> <ul class="su_nav-menu"> <li class="menu-toggle"> <i data-toggle="wy-nav-top" class="wy-nav-top-menu-button db-icon db-icon-menu pull-left"></i> <a href="https://www.databricks.com/" class="wy-nav-top-logo"><img src="../_static/small-scale-lockup-full-color-rgb.svg" width="137" height="21" alt="Databricks" /></a></li> <li><a href="https://help.databricks.com/s/">Help Center</a></li> <li class="active"> <a href="https://docs.databricks.com/en/index.html" >Documentation</a > </li> <li><a href="https://kb.databricks.com/">Knowledge Base</a></li> </ul> </div> <div class="su_nav-right"> <ul class="su_link-mobile">  </ul> <ul class="right-try-list"> <li><a href="https://community.databricks.com">Community</a></li> <li><a href="https://help.databricks.com">Support</a></li> <li><a href="mailto:doc-feedback@databricks.com?subject=Documentation Feedback">Feedback</a></li> <li class="try-databrick-item"><a href="https://signup.databricks.com/?dbx_source=docs">Try Databricks</a></li> </ul> </div> </nav> </nav> <div class="su_sub-header"> <div class="top-bar__container"> <div class="su_sub-header-inner top-bar"> <div class="su_subnav-menu-right"> <div id="auto" style="width: 100%;"> <div ng-controller="SearchautoController"> <div bind-html-compile="autocompleteHtml"> <form class="su__search-box-1" disabled="disabled"> <input class="su__search-input" type="search" name="Search box" id="su__search-b" placeholder="Search Documentation" disabled="disabled"/> <button class="su__search-button" type="submit" class="button button-success" disabled="disabled"> <svg width="24" height="24" viewBox="0 0 24 24"> <path d="M15.5 14h-.79l-.28-.27C15.41 12.59 16 11.11 16 9.5 16 5.91 13.09 3 9.5 3S3 5.91 3 9.5 5.91 16 9.5 16c1.61 0 3.09-.59 4.23-1.57l.27.28v.79l5 4.99L20.49 19l-4.99-5zm-6 0C7.01 14 5 11.99 5 9.5S7.01 5 9.5 5 14 7.01 14 9.5 11.99 14 9.5 14z" fill="#333"></path> </svg> </button> </form> </div> </div> </div> </div> <div class="top-bar__selectors"> <div class="translation-selector dropdown" translate="no"> <a href="javascript:void(0)" aria-haspopup="true"><img src="../_static/icons/globe.png" /><span class="translation-selector__current">English</span></a> <ul aria-label="submenu"> <li> <a href="../../en/mlflow/index.html" hreflang="en">English</a> </li> <li> <a href="../../ja/mlflow/index.html" hreflang="ja">日本語</a> </li> <li> <a href="../../pt/mlflow/index.html" hreflang="pt">Português</a> </li> </ul> </li> </div> <div class="cloud-selector dropdown" translate="no"> <a href="javascript:void(0)" aria-haspopup="true"><img src="../_static/icons/aws.svg" />Amazon Web Services</a> <ul aria-label="submenu"> <li> <a href='https://learn.microsoft.com/azure/databricks/mlflow/'><img src="../_static/icons/azure.svg" />Microsoft Azure</a> </li> <li> <a href="https://docs.gcp.databricks.com/mlflow/index.html"><img src="../_static/icons/gcp.svg" />Google Cloud Platform</a> </li> </ul> </li> </div> <script data-ot-ignore> const lang = ["en", "ja", "pt"]; const currentLanguage = window.location.pathname.split("/")[1]; const langIndex = lang.indexOf(currentLanguage); if (langIndex !== -1) { // Fixup cloud selector const languageSelector = document.querySelector(".translation-selector__current"); const languageLink = document.querySelector(`.translation-selector > ul a[hreflang="${currentLanguage}"]`); if (languageSelector && languageLink) { languageSelector.textContent = languageLink.textContent; languageLink.parentElement.parentElement.removeChild(languageLink.parentElement); } // Fixup links in cloud selector Array.from(document.querySelectorAll(`.cloud-selector > ul a`)).forEach(link => { link.pathname = `/${currentLanguage}${link.pathname}`; }); } </script> <script type="text/plain" class="optanon-category-C0003"> if (langIndex !== -1) { document.cookie = `lang=${currentLanguage}; expires=Fri, 31 Dec 9999 23:59:59 GMT;";` } </script> </div> </div> </div> </div> <page class="js-page-container"> <nav data-toggle="wy-nav-shift" class="wy-nav-side su_nav-side"> <div class="wy-side-scroll"> <div class="wy-side-nav-search"> </div> <div class="wy-menu wy-menu-vertical" data-spy="affix" role="navigation" aria-label="main navigation"> <a href="../index.html" class="main-navigation-home">Databricks on AWS</a> <p class="caption"><span class="caption-text">Get started</span></p> <ul> <li class="toctree-l1"><a class="reference internal" href="../getting-started/index.html">Get started</a></li> <li class="toctree-l1"><a class="reference internal" href="../introduction/index.html">What is Databricks?</a></li> <li class="toctree-l1"><a class="reference internal" href="../databricksiq/index.html">DatabricksIQ</a></li> <li class="toctree-l1"><a class="reference internal" href="../release-notes/index.html">Release notes</a></li> </ul> <p class="caption"><span class="caption-text">Load & manage data</span></p> <ul> <li class="toctree-l1"><a class="reference internal" href="../guides/index.html">Guides</a></li> <li class="toctree-l1"><a class="reference internal" href="../database-objects/index.html">Work with database objects</a></li> <li class="toctree-l1"><a class="reference internal" href="../connect/index.html">Connect to data sources</a></li> <li class="toctree-l1"><a class="reference internal" href="../compute/index.html">Connect to compute</a></li> <li class="toctree-l1"><a class="reference internal" href="../discover/index.html">Discover data</a></li> <li class="toctree-l1"><a class="reference internal" href="../query/index.html">Query data</a></li> <li class="toctree-l1"><a class="reference internal" href="../ingestion/index.html">Ingest data</a></li> <li class="toctree-l1"><a class="reference internal" href="../files/index.html">Work with files</a></li> <li class="toctree-l1"><a class="reference internal" href="../transform/index.html">Transform data</a></li> <li class="toctree-l1"><a class="reference internal" href="../jobs/index.html">Schedule and orchestrate workflows</a></li> <li class="toctree-l1"><a class="reference internal" href="../lakehouse-monitoring/index.html">Monitor data and AI assets</a></li> <li class="toctree-l1"><a class="reference internal" href="../external-access/index.html">Read with external systems</a></li> <li class="toctree-l1"><a class="reference internal" href="../data-sharing/index.html">Share data securely</a></li> </ul> <p class="caption"><span class="caption-text">Work with data</span></p> <ul class="current"> <li class="toctree-l1"><a class="reference internal" href="../delta-live-tables/index.html">Delta Live Tables</a></li> <li class="toctree-l1"><a class="reference internal" href="../structured-streaming/index.html">Structured Streaming</a></li> <li class="toctree-l1 current"><a class="reference internal" href="../machine-learning/index.html">AI and machine learning</a><ul class="current"> <li class="toctree-l2"><a class="reference internal" href="../machine-learning/ml-tutorials.html">Tutorials</a></li> <li class="toctree-l2"><a class="reference internal" href="../large-language-models/ai-playground.html">AI playground</a></li> <li class="toctree-l2"><a class="reference internal" href="../large-language-models/ai-functions.html">AI functions in SQL</a></li> <li class="toctree-l2"><a class="reference internal" href="../ai-gateway/index.html">AI Gateway</a></li> <li class="toctree-l2"><a class="reference internal" href="../machine-learning/model-serving/index.html">Deploy models</a></li> <li class="toctree-l2"><a class="reference internal" href="../machine-learning/train-model/index.html">Train models</a></li> <li class="toctree-l2"><a class="reference internal" href="../machine-learning/serve-data-ai.html">Serve data for AI</a></li> <li class="toctree-l2"><a class="reference internal" href="../generative-ai/agent-evaluation/index.html">Evaluate AI</a></li> <li class="toctree-l2"><a class="reference internal" href="../generative-ai/agent-framework/build-genai-apps.html">Build gen AI apps</a></li> <li class="toctree-l2"><a class="reference internal" href="../machine-learning/mlops/mlops-workflow.html">MLOps</a></li> <li class="toctree-l2 current"><a class="current reference internal" href="#">MLflow for AI agent and ML model lifecycle</a><ul> <li class="toctree-l3"><a class="reference internal" href="tracking.html">MLflow experiment tracking</a></li> <li class="toctree-l3"><a class="reference internal" href="../machine-learning/manage-model-lifecycle/index.html">Manage model lifecycle</a></li> </ul> </li> <li class="toctree-l2"><a class="reference internal" href="../machine-learning/retired-models-policy.html">Gen AI model maintenance policy</a></li> <li class="toctree-l2"><a class="reference internal" href="../machine-learning/integrations.html">Integrations</a></li> <li class="toctree-l2"><a class="reference internal" href="../machine-learning/graph-analysis.html">Graph and network analysis</a></li> <li class="toctree-l2"><a class="reference internal" href="../machine-learning/reference-solutions/index.html">Reference solutions</a></li> </ul> </li> <li class="toctree-l1"><a class="reference internal" href="../generative-ai/tutorials/ai-cookbook/index.html">Generative AI tutorial</a></li> <li class="toctree-l1"><a class="reference internal" href="../ai-bi/index.html">Business intelligence</a></li> <li class="toctree-l1"><a class="reference internal" href="../sql/index.html">Data warehousing</a></li> <li class="toctree-l1"><a class="reference internal" href="../notebooks/index.html">Notebooks</a></li> <li class="toctree-l1"><a class="reference internal" href="../delta/index.html">Delta Lake</a></li> <li class="toctree-l1"><a class="reference internal" href="../developers/index.html">Developers</a></li> <li class="toctree-l1"><a class="reference internal" href="../integrations/index.html">Technology partners</a></li> </ul> <p class="caption"><span class="caption-text">Administration</span></p> <ul> <li class="toctree-l1"><a class="reference internal" href="../admin/index.html">Account and workspace administration</a></li> <li class="toctree-l1"><a class="reference internal" href="../security/index.html">Security and compliance</a></li> <li class="toctree-l1"><a class="reference internal" href="../data-governance/index.html">Data governance (Unity Catalog)</a></li> <li class="toctree-l1"><a class="reference internal" href="../lakehouse-architecture/index.html">Lakehouse architecture</a></li> </ul> <p class="caption"><span class="caption-text">Reference & resources</span></p> <ul> <li class="toctree-l1"><a class="reference internal" href="../reference/api.html">Reference</a></li> <li class="toctree-l1"><a class="reference internal" href="../resources/index.html">Resources</a></li> <li class="toctree-l1"><a class="reference internal" href="../whats-coming.html">What’s coming?</a></li> <li class="toctree-l1"><a class="reference internal" href="../archive/index.html">Documentation archive</a></li> </ul> </div> <div role="contentinfo"> <p class="build_info notranslate"data-last-edit="January 13, 2025"> Updated Feb 18, 2025 </p> <script> window.addEventListener('DOMContentLoaded',function(){ var h1=document.querySelector('h1'); var bi=document.querySelector('[data-last-edit]'); if(h1 && bi){ var ver = document.createElement('p'); ver.className = 'version_info'; ver.textContent = bi.getAttribute('data-last-edit'); h1.parentElement.insertBefore(ver, h1.nextElementSibling); } }); </script> <p> <a id='feedbacklink' href="mailto:doc-feedback@databricks.com?subject=Documentation Feedback">Send us feedback</a> </p> </div> </div> </nav> <main class="wy-grid-for-nav su_nav-grid"> <section data-toggle="wy-nav-shift" class="wy-nav-content-wrap"> <div class="wy-nav-content su__nav_content"> <div class="rst-content"> <div role="navigation" aria-label="breadcrumbs navigation" class="wy-breadcrumbs-wrapper"> <ul class="wy-breadcrumbs"> <li><a href="../index.html">Documentation</a> <span class="db-icon db-icon-chevron-right"></span></li> <li><a href="../machine-learning/index.html">AI and machine learning on Databricks</a> <span class="db-icon db-icon-chevron-right"></span></li> <li>MLflow for gen AI agent and ML model lifecycle</li> <li class="wy-breadcrumbs-aside"> </li> </ul> </div> <div role="main" class="document" itemscope="itemscope" itemtype="http://schema.org/Article"> <div itemprop="articleBody"> <div class="section" id="mlflow-for-gen-ai-agent-and-ml-model-lifecycle"> <h1>MLflow for gen AI agent and ML model lifecycle<a class="headerlink" href="#mlflow-for-gen-ai-agent-and-ml-model-lifecycle" title="Permalink to this headline"> </a></h1> <p>This article describes how MLflow on Databricks is used to develop high-quality generative AI agents and machine learning models.</p> <div class="admonition note"> <p class="admonition-title">Note</p> <p>If you’re just getting started with Databricks, consider trying MLflow on <a class="reference internal" href="../getting-started/community-edition.html"><span class="doc">Databricks Community Edition</span></a>.</p> </div> <div class="section" id="what-is-mlflow"> <h2>What is MLflow?<a class="headerlink" href="#what-is-mlflow" title="Permalink to this headline"> </a></h2> <p>MLflow is an open source platform for developing models and generative AI applications. It has the following primary components:</p> <ul class="simple"> <li><p>Tracking: Allows you to track experiments to record and compare parameters and results.</p></li> <li><p>Models: Allow you to manage and deploy models from various ML libraries to various model serving and inference platforms.</p></li> <li><p>Model Registry: Allows you to manage the model deployment process from staging to production, with model versioning and annotation capabilities.</p></li> <li><p>AI agent evaluation and tracing: Allows you to develop high-quality AI agents by helping you compare, evaluate, and troubleshoot agents.</p></li> </ul> <p>MLflow supports <a class="reference external" href="https://www.mlflow.org/docs/latest/java_api/index.html">Java</a>, <a class="reference external" href="https://www.mlflow.org/docs/latest/python_api/index.html">Python</a>, <a class="reference external" href="https://www.mlflow.org/docs/latest/R-api.html">R</a>, and <a class="reference external" href="https://docs.databricks.com/api/workspace/experiments">REST</a> APIs.</p> </div> <div class="section" id="databricks-managed-mlflow"> <h2>Databricks-managed MLflow<a class="headerlink" href="#databricks-managed-mlflow" title="Permalink to this headline"> </a></h2> <p>Databricks provides a fully managed and hosted version of MLflow, building on the open source experience to make it more robust and scalable for enterprise use.</p> <p>The following diagram shows how Databricks integrates with MLflow to train and deploy machine learning models.</p> <div class="figure align-default"> <img alt="MLflow integrates with Databricks to manage the ML lifecycle." src="../_images/mlflow-databricks-integration-ml.png" /> </div> <p>Databricks-managed MLflow is built on Unity Catalog and the Cloud Data Lake to unify all your data and AI assets in the ML lifecycle:</p> <ol class="arabic simple"> <li><p><strong>Feature store:</strong> Databricks automated feature lookups simplifies integration and reduces mistakes.</p></li> <li><p><strong>Train models:</strong> Use Mosaic AI to train models or fine-tune foundation models.</p></li> <li><p><strong>Tracking</strong>: MLflow tracks training by logging parameters, metrics, and artifacts to evaluate and compare model performance.</p></li> <li><p><strong>Model Registry:</strong> MLflow Model Registry, integrated with Unity Catalog centralizes AI models and artifacts.</p></li> <li><p><strong>Model Serving:</strong> Mosaic AI Model Serving deploys models to a REST API endpoint.</p></li> <li><p><strong>Monitoring:</strong> Mosaic AI Model Serving automatically captures requests and responses to monitor and debug models. MLflow augments this data with trace data for each request.</p></li> </ol> </div> <div class="section" id="model-training"> <h2>Model training<a class="headerlink" href="#model-training" title="Permalink to this headline"> </a></h2> <p>MLflow Models are at the core of AI and ML development on Databricks. MLflow Models are a standardized format for packaging machine learning models and generative AI agents. The standardized format ensures that models and agents can be used by downstream tools and workflows on Databricks.</p> <ul class="simple"> <li><p>MLflow documentation - <a class="reference external" href="https://mlflow.org/docs/latest/models.html">Models</a>.</p></li> </ul> <p>Databricks provides features to help you train different kinds of ML models.</p> <ul class="simple"> <li><p><a class="reference internal" href="../machine-learning/train-model/index.html"><span class="doc">Train AI models using Mosaic AI</span></a>.</p></li> </ul> </div> <div class="section" id="experiment-tracking"> <h2>Experiment tracking<a class="headerlink" href="#experiment-tracking" title="Permalink to this headline"> </a></h2> <p>Databricks uses MLflow experiments as organizational units to track your work while developing models.</p> <p>Experiment tracking lets you log and manage parameters, metrics, artifacts, and code versions during machine learning training and agent development. Organizing logs into experiments and runs allows you to compare models, analyze performance, and iterate more easily.</p> <ul class="simple"> <li><p><a class="reference internal" href="tracking.html"><span class="doc">Experiment tracking using Databricks</span></a>.</p></li> <li><p>See MLflow documentation for general information on <a class="reference external" href="https://mlflow.org/docs/latest/tracking.html">runs and experiment tracking</a>.</p></li> </ul> </div> <div class="section" id="model-registry-with-unity-catalog"> <h2>Model Registry with Unity Catalog<a class="headerlink" href="#model-registry-with-unity-catalog" title="Permalink to this headline"> </a></h2> <p>MLflow Model Registry is a centralized model repository, UI, and set of APIs for managing the model deployment process.</p> <p>Databricks integrates Model Registry with Unity Catalog to provide centralized governance for models. Unity Catalog integration allows you to access models across workspaces, track model lineage, and discover models for reuse.</p> <ul class="simple"> <li><p><a class="reference internal" href="../machine-learning/manage-model-lifecycle/index.html"><span class="doc">Manage models using Databricks Unity Catalog</span></a>.</p></li> <li><p>See MLflow documentation for general information on <a class="reference external" href="https://mlflow.org/docs/latest/model-registry.html">Model Registry</a>.</p></li> </ul> </div> <div class="section" id="model-serving"> <h2>Model Serving<a class="headerlink" href="#model-serving" title="Permalink to this headline"> </a></h2> <p>Databricks Model Serving is tightly integrated with MLflow Model Registry and provides a unified, scalable interface for deploying, governing, and querying AI models. Each model you serve is available as a REST API that you can integrate into web or client applications.</p> <p>While they are distinct components, Model Serving heavily relies on MLflow Model Registry to handle model versioning, dependency management, validation, and governance.</p> <ul class="simple"> <li><p><a class="reference internal" href="../machine-learning/model-serving/index.html"><span class="doc">Model Serving using Databricks</span></a>.</p></li> </ul> </div> <div class="section" id="ai-agent-development-and-evaluation"> <span id="model-serving"></span><h2>AI agent development and evaluation<a class="headerlink" href="#ai-agent-development-and-evaluation" title="Permalink to this headline"> </a></h2> <p>For AI agent development, Databricks integrates with MLflow similarly to ML model development. However, there are a few key differences:</p> <ul class="simple"> <li><p>To create AI agents on Databricks, use <a class="reference internal" href="../generative-ai/agent-framework/build-genai-apps.html"><span class="doc">Mosaic AI Agent Framework</span></a>, which relies on MLflow to track agent code, performance metrics, and agent traces.</p></li> <li><p>To evaluate agents on Databricks, use <a class="reference internal" href="../generative-ai/agent-evaluation/index.html"><span class="doc">Mosaic AI Agent Evaluation</span></a>, which relies on MLflow to track evaluation results.</p></li> <li><p>MLflow tracking for agents also includes <a class="reference external" href="https://mlflow.org/docs/latest/llms/tracing/index.html">MLflow Tracing</a>. MLflow Tracing allows you to see detailed information about the execution of your agent’s services. Tracing records the inputs, outputs, and metadata associated with each intermediate step of a request, letting you quickly find the source of unexpected behavior in agents.</p></li> </ul> <p>The following diagram shows how Databricks integrates with MLflow to create and deploy AI agents.</p> <div class="figure align-default"> <img alt="MLflow integrates with Databricks to manage the genAI app lifecycle." src="../_images/mlflow-databricks-integration-agents.png" /> </div> <p>Databricks-managed MLflow is built on Unity Catalog and the Cloud Data Lake to unify all your data and AI assets in the genAI app lifecycle:</p> <ol class="arabic simple"> <li><p><strong>Vector & feature store:</strong> Databricks automated vector and feature lookups simplify integration and reduce mistakes.</p></li> <li><p><strong>Create and evaluate AI agents:</strong> Mosaic AI Agent Framework and Agent Evaluation help you create agents and evaluate their output.</p></li> <li><p><strong>Tracking & tracing:</strong> MLflow tracing captures detailed agent execution information for enhanced genAI observability.</p></li> <li><p><strong>Model Registry:</strong> MLflow Model Registry, integrated with Unity Catalog centralizes AI models and artifacts.</p></li> <li><p><strong>Model Serving:</strong> Mosaic AI Model Serving deploys models to a REST API endpoint.</p></li> <li><p><strong>Monitoring:</strong> MLflow automatically captures requests and responses to monitor and debug models.</p></li> </ol> </div> <div class="section" id="open-source-vs-databricks-managed-mlflow-features"> <h2>Open source vs. Databricks-managed MLflow features<a class="headerlink" href="#open-source-vs-databricks-managed-mlflow-features" title="Permalink to this headline"> </a></h2> <p>For general MLflow concepts, APIs, and features shared between open source and Databricks-managed versions, refer to <a class="reference external" href="https://mlflow.org/docs/latest/index.html">MLflow documentation</a>. For features exclusive to Databricks-managed MLflow, see Databricks documentation.</p> <p>The following table highlights the key differences between open source MLflow and Databricks-managed MLflow and provides documentation links to help you learn more:</p> <table class="docutils align-default"> <colgroup> <col style="width: 33%" /> <col style="width: 33%" /> <col style="width: 33%" /> </colgroup> <thead> <tr class="row-odd"><th class="head"><p>Feature</p></th> <th class="head"><p>Availability on open source MLflow</p></th> <th class="head"><p>Availability on Databricks-managed MLflow</p></th> </tr> </thead> <tbody> <tr class="row-even"><td><p>Security</p></td> <td><p>User must provide their own security governance layer</p></td> <td><p><a class="reference internal" href="../security/index.html"><span class="doc">Databricks enterprise-grade security</span></a></p></td> </tr> <tr class="row-odd"><td><p>Disaster recovery</p></td> <td><p>Unavailable</p></td> <td><p><a class="reference internal" href="../admin/disaster-recovery.html"><span class="doc">Databricks disaster recovery</span></a></p></td> </tr> <tr class="row-even"><td><p>Experiment tracking</p></td> <td><p><a class="reference external" href="https://mlflow.org/docs/latest/tracking.html">MLflow Tracking API</a></p></td> <td><p>MLflow Tracking API integrated with <a class="reference internal" href="tracking.html"><span class="doc">Databricks advanced experiment tracking</span></a></p></td> </tr> <tr class="row-odd"><td><p>Model Registry</p></td> <td><p><a class="reference external" href="https://mlflow.org/docs/latest/model-registry.html">MLflow Model Registry</a></p></td> <td><p><a class="reference internal" href="../machine-learning/manage-model-lifecycle/index.html"><span class="doc">MLflow Model Registry integrated with Databricks Unity Catalog</span></a></p></td> </tr> <tr class="row-even"><td><p>Unity Catalog integration</p></td> <td><p>Open source integration with Unity Catalog</p></td> <td><p><a class="reference internal" href="../machine-learning/manage-model-lifecycle/index.html"><span class="doc">Databricks Unity Catalog</span></a></p></td> </tr> <tr class="row-odd"><td><p>Model deployment</p></td> <td><p>User-configured integrations with external serving solutions (SageMaker, Kubernetes, container services, and so on)</p></td> <td><p><a class="reference internal" href="../machine-learning/model-serving/index.html"><span class="doc">Databricks Model Serving</span></a> and external serving solutions</p></td> </tr> <tr class="row-even"><td><p>AI agents</p></td> <td><p><a class="reference external" href="https://mlflow.org/docs/2.10.2/llms/index.html">MLflow LLM development</a></p></td> <td><p>MLflow LLM development integrated with <a class="reference internal" href="../generative-ai/agent-framework/build-genai-apps.html"><span class="doc">Mosaic AI Agent Framework</span></a> and <a class="reference internal" href="../generative-ai/agent-evaluation/index.html"><span class="doc">Agent Evaluation</span></a></p></td> </tr> <tr class="row-odd"><td><p>Encryption</p></td> <td><p>Unavailable</p></td> <td><p>Encryption using <a class="reference internal" href="../security/keys/index.html"><span class="doc">customer-managed keys</span></a></p></td> </tr> </tbody> </table> <div class="toctree-wrapper compound"> </div> </div> </div> </div> </div> <div class="suapp-rating"> <div id="suPageRateApp"> <su-app></su-app> </div> </div> <hr> <footer> <div role="contentinfo"> <p class="copyright"> © Databricks 2025. All rights reserved. Apache, Apache Spark, Spark, and the Spark logo are trademarks of the <a href="http://www.apache.org/">Apache Software Foundation</a>. </p> <p> <a id='feedbacklink' href="mailto:doc-feedback@databricks.com?subject=Documentation Feedback">Send us feedback</a> | <a href="https://www.databricks.com/legal/privacynotice">Privacy Notice</a> | <a href="https://www.databricks.com/terms-of-use">Terms of Use</a> | <a href="https://www.databricks.com/legal/modern-slavery-policy-statement">Modern Slavery Statement</a> | <a href="https://www.databricks.com/legal/supplemental-privacy-notice-california-residents">California Privacy</a> | <a href="javascript: OneTrust.ToggleInfoDisplay()">Your Privacy Choices</a> <img alt="" src="https://www.databricks.com/sites/default/files/2022-12/gpcicon_small.png" data-ot-ignore="1" class="dbgpcicon inline-block max-h-[15px] w-auto" width="31" height="15"> </p> </div> </footer> </div> </div> </section> </main> </page> </body> </html>