Meet Managed Service for MLflow in general availability

Today, we’re happy to announce that Managed Service for MLflow became publicly available for everyone. Since it was launched in preview last September, we’ve made some improvements in its stability, to ensure a smooth user experience. By combining core functionality for MLOps with the easy-to-use cloud deployment, this service helps data scientists and ML engineers to simplify and accelerate the model development process significantly.

Essential product for ML professionalsEssential product for ML professionals

MLflow is a popular and widely used MLOps software that enables ML professionals to track model runs and experiments, streamline model management and enhance cross-team collaboration. Using this MLOps tool is necessary to ensure model reproducibility and metadata consistency when developing all kinds of projects: from small task-oriented ML workflows to large-scale generative AI agents.

MLflow running in the cloudMLflow running in the cloud

At Nebius, we believe that democratizing AI will eventually benefit all industry players and their customers. For that reason, we continue developing our ML tools as managed, cloud-based services with minimal effort needed for infrastructure deployment and maintenance.

The managed service for MLflow delivers an experience that feels like SaaS — you simply log into the console, set up the MLflow cluster, start collecting your training metadata and begin interacting with model artefacts via the MLflow UI.

Figure 2. Creating an MLflow cluster in the Nebius cloud console

Figure 3. MLflow clusters are launched in the Nebius cloud console

Figure 4. Connect a runtime environment to the MLflow cluster

Figure 5. Using the MLflow UI to track training runs