sub2sub2 / AI-LINUX-RUST

0 stars 2 forks source link

Extensibility Issue for App #15

Open sseoreo opened 4 months ago

sseoreo commented 4 months ago

I have thought of some things about our system, which are as follows:

  1. In the view of App developers, how can they know the provided data or models from MCA when to develop their app within our system?
  2. In the view of MCA, how can we provide data to App developers while maintaining extensibility and security?
sseoreo commented 4 months ago

I suggest a scenario for this issue.

In our system, AI models are supposed to be developed prior to apps. Therefore, how bout we assume that apps will be developed or registered after AI models?

Under the assumption,

  1. At the beginning of this system, we have MCA.
  2. Next, the first AI model will be registered on MCA, and then MCA will define APIs based on I/O types of the AI model. The APIs are implemented as gRPC services and released to app developers.
  3. Last, app developers will use the APIs in their own apps

In this scenario, MCA has 2 types of functions:

  1. Based on IO types of AI models, MCA builds gRPC services
  2. MCA manages data flows (or access points) from apps to AI models

@hyunsube How bout that?