Verta.ai ModelDB on Amazon Web Services
ModelDB is an end-to-end system for tracking, versioning and auditing machine learning models. It ingests models and associated metadata as models are being trained, stores model data in a structured format, and surfaces it through a web-frontend for rich querying and the python client.
Visit ModelDB code repository for more details.
The vision for this project is to develop AI Platform capabilities integrated with Amazon SageMaker. Amazon SageMaker currently comes with its own experiment tracking, model versioning features. However, data versioning is not yet available. We have added examples notebooks to demonstrate how to integrate ModelDB data versioning into your Amazon SageMaker workflow.
In addition, we are working on full integration with Amazon SageMaker Experiments API and/or training logs so as to provide a unified experience in machine learning workflow. We do not have a defined roadmap yet, but we hope the project matures as it simplifies machine learning workflows.
This project is deployed via Amazon Cloudformation using the AWS Cloud Development Kit. Follow the steps below to deploy the stack in your AWS Account.
$ sudo npm install -g aws-cdk
$ cd stack
$ npm install -s
$ cdk deploy -y
The stack creates the following AWS resources:
After a successful deployment, you should be able to access ModelDB frontend UI.