swe1010/CICD-MLOPS-Boston-Housing

MLOps CI/CD — GitHub Actions + Hugging Face Hub

Every push to main trains a model, evaluates it, and — only if it passes an accuracy threshold — deploys it to Hugging Face Hub automatically.

How it works

push/PR to GitHub
        │
        ▼
  ┌──────┐
  │ test │   pytest sanity checks
  └──┬───┘
     ▼
  ┌────────────────────┐
  │ train_and_evaluate  │  prepare → train → evaluate
  └──────┬──────────────┘  ← fails here if accuracy < params.yaml threshold
         ▼
  ┌─────────────────────┐
  │ deploy_to_huggingface │  only on push to main, only if gate passed
  └─────────────────────┘  re-trains + pushes model to your HF repo

Setup

1. Create a Hugging Face account + token

• Go to https://huggingface.co/settings/tokens
• Create a token with Write access
• Copy it — you'll paste it into GitHub next

2. Create the GitHub repo and push this code

cd mlops-simple
git init
git add .
git commit -m "Initial MLOps pipeline"
git branch -M main
git remote add origin https://github.com/<your-username>/<your-repo>.git
git push -u origin main

3. Add your HF token as a GitHub secret

In your GitHub repo: Settings → Secrets and variables → Actions → New repository secret

• Name: HF_TOKEN
• Value: the token from step 1

4. Add your target HF repo as a GitHub variable

Same page, Variables tab → New repository variable

• Name: HF_REPO_ID
• Value: your-hf-username/your-model-name (e.g. mupalaniappan/breast-cancer-rf) — it'll be created automatically on first successful run if it doesn't exist yet.

5. Push to main

That's it — the workflow runs automatically. Check the Actions tab in GitHub to watch it, and once deploy_to_huggingface finishes, your model is live at https://huggingface.co/your-hf-username/your-model-name.

Adjusting the quality gate

Edit evaluate.min_accuracy in params.yaml. Anything scoring below that never reaches Hugging Face — the deploy_to_huggingface job simply doesn't run.

Try it locally first (optional but recommended)

pip install -r requirements.txt
python src/prepare.py
python src/train.py
python src/evaluate.py
cat metrics.json

CI/CD deployment configured.