This project consists of AI App, Model Control Agent(MCA), and AI model.
-
Intall and set docker in Mac (refer. https://docs.github.com/en/packages/working-with-a-github-packages-registry/working-with-the-container-registry#authenticating-to-the-container-registry)
-
Pull and run the docker image
$docker pull ghcr.io/sub2sub2/ai_rust:1.0.0
$docker run -it --name AI_RUST ghcr.io/sub2sub2/ai_rust:1.0.0 /bash/bin
- Access into the running container and clone the git
$docket attach AI_RUST
--- in docker shell ---
clone the git repo and test
AI App is an app that needs AI inference to anaylze any type of data supported by MCA. The AI app can request AI model inference to MCA and get a result of the inference. App developers can access to MCA through on gRPC and request an inference to MCA.
Model Control Agent (MCA) is an agent that handles AI model inference requested by AI App. The agent has predefined AI model information, so it connects AI App to AI Model by D-bus which is the linux IPC system. Also, MCA has control logic for the inference and each model. Server maintainers, app developers, and model engineers can register their own model easily with MCA.
AI Models are AI models handle inference requested by MCA. The models will be launched on demend way. AI Models may get file path that indicates the coming file from AI app or actual data with json string form. You need to train a model and convert it to onnx model.