DEVOPSdigest

Iterative launched Machine Learning Engineering Management (MLEM) – an open source model deployment and registry tool that uses an organization’s existing Git infrastructure and workflows.

MLEM bridges the gap between ML engineers and DevOps teams. DevOps teams can easily understand the underlying frameworks and libraries a model uses and automate deployment into a one-step process for production services and apps.

“Iterative enables customers to treat AI models as just another type of software artifact,” said Sriram Subramanian, research director, AI/ ML Lifecycle Management Software, IDC. “The ability to build ML model registries using Git infrastructure and DevOps principles allows models to get into production faster.”

MLEM is a core building block for a Git-based ML model registry, together with other Iterative tools, like GTO and DVC. A model registry stores and versions trained ML models. Model registries greatly simplify the task of tracking models as they move through the ML lifecycle, from training to production deployments and ultimately retirement.

“Model registries simplify tracking models moving through the ML lifecycle by storing and versioning trained models, but organizations building these registries end up with two different tech stacks for machine learning models and software development,” said Dmitry Petrov, co-founder and CEO of Iterative. “MLEM as a building block for model registries uses Git and traditional CI/CD tools, aligning ML and software teams so they can get models into production faster.”

With Iterative tools, organizations can build a ML model registry based on software development tools and best practices. This means Git acts as a central source of truth for models, eliminating the need for external tools specific to machine learning. All information around a model including which are in production, development, or deprecated, can all be viewed in Git.

MLEM’s modular nature fits into any organization’s software development workflows based on Git and CI/CD, without engineers having to transition to a separate machine learning deployment and registry tool. This allows teams to use a similar process across both ML models and applications for deployment, eliminating duplication in processes and code. Teams are then able build a model registry in hours rather than days.

MLEM promotes a comprehensive machine learning model lifecycle management workflow using a GitOps-based approach. Software development and MLOps teams can then be aligned, using the same tools to speed the time it takes a model to get from development to production.