June 23, 2025: Today, Canonical announced the release of Charmed Feast, an enterprise solution for feature management with seamless integration with Charmed Kubeflow, Canonical’s distribution of the popular open source MLOps platform. Charmed Feast provides the full breadth of the upstream Feast capabilities, adding multi-cloud capabilities, and comprehensive support.Feast is an open source operational data system for managing and serving machine learning (ML) features to models during both training and inference. It acts as a bridge between data engineering and machine learning, enabling consistent access to feature data in real-time and batch environments. Feast simplifies the process of building, versioning, and deploying features so teams can reuse features across workflows and reduce duplication of effort which helps to scale their machine learning initiatives more efficiently and deliver intelligent applications faster and with greater consistency.Feature stores play a critical role in the AI/ML lifecycle. From model development and fine tuning to serving data in production for inference—a Feature store is a critical tool for production AI. Feast bridges the gap between Data Scientists, Data Engineers, MLOps Engineers, and Software Engineers to give users tooling to take their software to production. Building on its deep integration with Kubeflow, Feast is now charting an even brighter future with investments in generative AI and retrieval-augmented generation (RAG).Francisco Javier ArceoFeast Maintainer & Kubeflow Steering Committee memberCharmed Feast is a fully supported, production-grade distribution of Feast. Designed to seamlessly integrate with Charmed Kubeflow and the rest of Canonical’s MLOps and big data ecosystem (which also includes MLFlow, OpenSearch, and Spark), Charmed Feast simplifies feature engineering and delivery, enabling teams to reproducibly manage features and serve them to models in production with ease. With Charmed Feast, organizations get the power of open source, backed by Canonical’s operational excellence and security guarantees, with long-term support, regular security updates, and optional 24/7 enterprise-grade SLAs.Built-in integration with Charmed Kubeflow and the Canonical MLOps ecosystemEffortless deployment and seamless integration with Charmed Kubeflow enables teams to serve consistent features to both training and production environments within Kubeflow pipelines, minimize retraining inefficiencies and prevent performance drops. Charmed Feast can run on the same Kubernetes cluster as Charmed Kubeflow, serving as a unified source of feature data for pipelines and model serving. Data teams can manage, version, and serve features from a single platform, reducing training-serving skew and accelerating deployment cycles while maintaining reliability.Operationalizing ML pipelines at scaleAs a new addition to Canonical’s growing MLOps and big data portfolio , Charmed Feast helps unify and operationalize the full machine learning lifecycle. Charmed Feast is delivered as a Juju charm (charmed operator), making installation and management highly automated and portable. With a single Juju command, teams can deploy Charmed Feast on any infrastructure – from the public cloud to an on-premises Kubernetes cluster. This model-driven approach simplifies scaling and lifecycle management: the charm encapsulates best practices for configuring Feast, handling updates, and integrating with other services. As a result, data engineers get a consistent deployment experience across environments, and can easily move feature store workloads between clouds or data centers without retooling.Strengthening the open source communityFeast has long been a first-class add-on in the Kubeflow ecosystem, with both communities working closely to ensure a smooth and robust integration experience for users. Canonical actively contributes to the Kubeflow project and maintains a strong relationship with the Feast community. By packaging Feast as a charm and including it in our supported MLOps platform, we help amplify upstream innovation while making it more accessible and enterprise-ready. This approach enables users to adopt open standards confidently, with the peace of mind that comes from Canonical’s engineering support.Additional benefits of Charmed FeastSimple, per-node pricing: Charmed Feast is part of Canonical’s data and AI portfolio. Customers can subscribe to enterprise support through Ubuntu Pro + Support on a predictable per-node basis. This subscription covers the full suite of Canonical’s charmed applications – including Charmed Kubeflow, MLFlow, Spark, OpenSearch, PostgreSQL, MongoDB, and Kafka – without additional software license fees. This makes budgeting and financial planning straightforward, while giving teams the freedom to deploy integrated solutions like Charmed Feast at no extra cost.Long-term security and support: Canonical provides up to 10 years of security maintenance and CVE patching for Charmed Feast. Optional 24/7 enterprise support ensures high availability and stability for production workloads, so you can confidently operate mission-critical feature management services.Fully managed option available: Canonical’s Managed MLOps service offers automation, scalability, availability, observability integration, and hands-on support from trusted experts. This option helps teams reduce operational complexity and concentrate on building data-driven applications instead of maintaining infrastructure.Start building with Charmed Feast today With Charmed Feast, you can bring order to your feature engineering workflows, reduce inconsistencies across environments, and deploy ML pipelines with greater speed and reliability. Whether you’re just getting started or scaling enterprise workloads, Charmed Feast offers the right balance of open source flexibility and production-grade assurance.To get started with Charmed Feast, refer to the documentation. For more information, visit https://canonical.com/mlops/feast. About CanonicalCanonical, the publisher of Ubuntu, provides open source security, support and services. Our portfolio covers critical systems, from the smallest devices to the largest clouds, from the kernel to containers, from databases to AI. With customers that include top tech brands, emerging startups, governments and home users, Canonical delivers trusted open source for everyone. Learn more at https://canonical.com/