Introduction

Goal is to deploy a simple ML workload (object detection) and scale it using kubernetes

Getting started with development

installation

# make new venv
python -m venv E:\venvs\ml-kubernetes
# activate venv
E:\venvs\ml-kubernetes\Scripts\activate.bat

running in dev

cd api && uvicorn server:app --reload

runnning tests

cd api && pytest

Notes

dependencies

• fastapi
• "uvicorn[standard]" # server
• httpx # http client for testing
• pytest # test framework
• pytest-asyncio # for async functions
• opencv # for yolo
• requests # for downloading model
• python-multipart # for file upload handling

dev dependencies

• flake8 # code linting
• black # code formatting

useful commands

cd api && pip freeze > requirements.txt

Next steps

• learn more about python modules
  • how init.py helps?
  • different b/w relative and abs imports