ml-eng-lab System Architecture

Notebook experiments, local runtimes, verification, and published documentation surfaces.
ml-eng-lab system architecture Architecture diagram showing contributors, runtime environments, notebook tasks, validation, CI, dependencies, and documentation publishing. Contributor boundary Maintainer edits + runs notebooks Runtime environments Execution Shells venv, Docker, Codespaces genai-vanilla JupyterHub Repository core notebooks/ 21 tasks, 29 active notebooks local ./data + ./runs Read-only archive notebooks/archive CodeXGLUE history Makefile papermill tier targets cd into notebook dirs Verifier + Tests scripts/verify_repo.py pytest + ruff CI Workflows Tier-A, smoke tiers Docker + docs build Docs Source README + docs/ MkDocs Material GitHub Pages published docs site ml-eng-lab URL External Dependencies thekaveh-nnx, Torch, PyG spaCy + NLTK assets Legend: runtime / validation repository data automation / CI published documentation external or historical

Notebook Boundary

  • Active work runs from `notebooks/<task>/`.
  • `./data` and `./runs` stay notebook-local.
  • Archive content remains read-only.

Validation Boundary

  • Make targets preserve notebook CWD semantics.
  • Verifier, pytest, and ruff cover local gates.
  • CI replays the same command surfaces.

Documentation Boundary

  • README is the repository entry point.
  • MkDocs publishes focused docs pages.
  • Pages deploys from the checked-in docs tree.