full-stack-ml

Demo full stack machine learning

Developer note

Local prerequisites

• Linux, MacOS or Windows with WSL 2:
• Python >=3.9: Python Docs
• Node: Node Docs
• Pip: Pip Docs
• Git: Git Docs
• pre-commit: Pre-commit Docs
• Poetry: Poetry Docs
• Docker and Docker Compose:
  • Docker Engine Docs
  • Docker Compose Docs
• Snyk: Snyk (Requires license)

Add libraries and projects as source root:

• projects/backend
• libraries/imagenet
• etc.

Run with Docker Compose

Make sure .env is populated, then run docker-compose:

docker-compose up

"Code" of conduct

We apply the following philosophies:

• Pre-commit with black, flake8, isort and prettier to ensure uniform style throughout the project
• Pre-commit with bandit to uncover security issues within the code
• Conventional commits
• Semantic versioning

Todo:

• Add .snyk for dependency scanning
• Add GitHub Actions pipeline for CI and later CD.
• Create explicitly separate network for backend network vs. Traefik proxy. Ensure we get certificates.
• Ensure docker containers are running as non-root -> bitnami containers or manual handling
• Use KeyVault or similar for secrets handling instead of environmental variables
• Set up test db.
• Implement OpenAPI Generator into the CI/CD
• Run Imagenet/Tensorflow as a separate service.

Long term todo:

• Change to Oauth (Authorization flow)[https://docs.microsoft.com/en-us/azure/active-directory/develop/v2-oauth2-auth-code-flow] or other flow depending on needs.
• Use OAuth2 scopes/RBAC such as (Azure AD / Identity Platform)[https://docs.microsoft.com/en-us/azure/active-directory/develop/v2-permissions-and-consent]
  • See (FastAPI OAuth2 scops details)[https://fastapi.tiangolo.com/advanced/security/oauth2-scopes/]

Features:

• RBAC
• Use OpenAPI spec for frontend
• Axio for API handling
• Implement machine learning for logged in users