Verta

Verta

Standalone platform for Model Management & Operations

Commercial software for on-premise or private cloud deployment
Managed service in customer cloud or Verta hosted

On-premise (air-gapped)
Private/public cloud platform
Verta hosted offering

Seat-based license

SLOs/SLAs are guranteed and can be customized per contract agreement

24x7 support is available based on the contract agreement

RBAC and user management system along with SSO support to integrate into identity systems (e.g SAML, Okta, Active Directory, etc.)

Advanced user management, SSO, RBAC, integrations with enterprise identity providers, audit/access logs, model release approval workflows

Verta hosted - You can get started within minutes (sign-up with the platform, setup our client library and start logging model metadata).
Verta managed cloud deployment - We deploy and push upgrades to customer cloud
Onprem, VPC and cloud deployment - We provide helm charts for customers to quickly deploy. Production deployments varies by the complexity of the infrastructure

The web UI and APIs are designed for scalability and flexibility. You can log metadata and data types to fit your use case. With open APIs you access models and metadata from Verta and push them to other tools.

You can package and log model API and all the artifacts in Verta Model Registry and Experiment Management system
You can capture and track code version, data versions, environment dependencies, metrics and hyper-params for model reproducibility

Code versions, commit SHA, notebook snapshots
Dataset versions and distribution
Metrics
Hyperparameters
Observations (recurring metadata that are repeatedly measured over time - e.g. batch loss over epoch and memory usage)
Attributes (rich are descriptive metadata about your model performance, model type, dataset feature distribution, tables, charts, histograms and other information
Labels and tags
Environment dependencies and requirements

Compare selected experiments with a tabular view (compare all the artifacts, code, dataset and model metadata). Includes charts to visualize and compare metrics, hyper-params and attributes.
Compare model version - In Verta Model Registry, compare different model versions across their lifecycle in an easy to use tabular interface (compare model test results, datasets, artifacts, code versions and more)
Dashboard and charts to compare and group experiment runs (when you are comparing and looking at trends across many runs)

Manage users and teams and organizations with workspaces
Projects, Experiments and Experiment Runs help group experiments and models
Experiment runs can be searched and filtered based on tags, names and experiments (groups)
With interactive charts and dashboards you can build custom views and share with your teams and peers
With a robust access control in place, you can add collaborators and decide what they can see and act on

Verta model registry supports appoval workflows and you can transition models from development, staging or production and retirement. Customers can define the right roles and permission for who can approve a stage change or deploy a model. You can add comments, share custom dashboards for collaboration and transparency across teams.

Activities are logged, providing signals and tracking for updates and promotions across development stages.

We gurantee immurability for model artifacts, requirements and attributes for compliance and safety. Admins can choose from 3 different lock levels (closed, redact and open) based on the company policy and model types.

Through open APIs and eventing ststems Verta model registry provides CI/CD automation workflows with CI/CD tools like Jenkins.

Verta integrates with popular ML libraries and has deep integrations with several DevOps tools like Datadog, Kafka, Github and Jenkins