3. RequirementsΒΆ

As described on the Acumos.org website, Acumos AI is a platform and open source framework that makes it easy to build, share, and deploy AI apps, and operate the Acumos portals that enable those capabilities. Acumos standardizes the infrastructure stack and components required to run an out-of-the-box general AI environment. This frees data scientists and model trainers to focus on their core competencies and accelerates innovation.

The Acumos platform enables the following high-level set of capabilities in support of the goals above, which are fulfilled through the various components and interfaces of the Acumos platform architecture:

  • Build machine-learning models and solutions
    • Use client libraries to generate model package for onboarding by CLI or Web
    • Generate model microservice images with embedded model runners based upon an Ubuntu docker base image
    • Design and generate composite solutions as a directed graph of multiple model microservices, with additional supporting components
  • Share models and solutions
    • Onboard models by CLI and Web
    • Share with your team, and publish to company and public marketplaces
    • Federate multiple Acumos portals for model/solution distribution
  • Deploy models and solutions
    • Download for local deployment under docker and kubernetes
    • Deploy to public and private clouds (Azure, OpenStack)
    • Interact with models, and observe solution-internal dataflow
  • Operate Acumos platforms
    • Deploy the platform under docker or kubernetes, as a single-node (all-in-one) or multi-node platform
    • Secure the platform
    • Administer the platform via the portal UI
    • logging and analytics collectiom, storage, and visualization