/MLOps-SOAI-Workshop

Slides, code and demo of some CD/CI tasks for ML models deployed using MLFlow + Streamlit + FastAPI

Primary LanguagePython

MLOps workshop with MLflow|Streamlit|FastAPI|python

Outline

  • What's MLOps and why need CD/CI ops in AI projects ?
  • Presenting MLflow for ML models lifecycle management
  • Proof-of-concept of the robustness of MLflow+FastAPI+Steamlit deployement stack
  • End-to-end implementation and Demo tutorial
  • Application on managing a simple torch Linear model for digit recognition

How to launch the scripts

  1. cd to Code folder
  2. In a DB Browser open sqlite database 'test.db'
  3. Execute the following commands in seperate CLIs
    • "streamlit run frontend/streamlit_dashboard.py"
    • "uvicorn backend.main:app"
    • "mlflow ui --backend-store-uri sqlite:///test.db"

Proposed dev stack