UPLOAD_GUIDE.md

File Upload Guide: Where Each File Goes

This guide shows exactly which files are uploaded to which location (Hugging Face or GitHub) and when.

---

Overview: Three Upload Destinations

1. Hugging Face Dataset Repo (ananttripathiak/engine-maintenance-dataset)
2. Hugging Face Model Repo (ananttripathiak/engine-maintenance-model)
3. Hugging Face Space (ananttripathiak/engine-maintenance-space)
4. GitHub Repository (ananttripathi/engine-predictive-maintenance)

---

1. Hugging Face Dataset Repo

Repo ID: ananttripathiak/engine-maintenance-dataset
Created by: src/data_register.py and src/data_prep.py

Files Uploaded:

A. Raw Data (via src/data_register.py)

• File: data/engine_data.csv
• Uploaded to: data/engine_data.csv in the dataset repo
• When: Run python src/data_register.py

B. Processed Data (via src/data_prep.py)

• File: data/processed/train.csv

• Uploaded to: data/train.csv in the dataset repo

• When: Run python src/data_prep.py

• File: data/processed/test.csv

• Uploaded to: data/test.csv in the dataset repo

• When: Run python src/data_prep.py

Scripts that upload here:

• src/data_register.py → uploads raw data
• src/data_prep.py → uploads train/test splits

---

2. Hugging Face Model Repo

Repo ID: ananttripathiak/engine-maintenance-model
Created by: src/train.py

Files Uploaded:

• File: models/best_model.joblib
• Uploaded to: model.joblib in the model repo
• When: Run python src/train.py (after training completes)

Scripts that upload here:

• src/train.py → uploads the trained model

---

3. Hugging Face Space (Streamlit App)

Repo ID: ananttripathiak/engine-maintenance-space
Created by: src/deploy_to_hf.py

Files Uploaded:

The src/deploy_to_hf.py script uploads the entire project folder except:

• data/ (ignored - too large)
• mlruns/ (ignored - MLflow tracking data)
• models/ (ignored - model is in model repo)
• .github/ (ignored - GitHub-specific)

Files that ARE uploaded to Space:

• src/app.py ← Main Streamlit app
• src/inference.py ← Inference utilities
• src/config.py ← Configuration
• Dockerfile ← Container definition
• requirements.txt ← Python dependencies
• README.md ← Documentation
• Other src/*.py files (if needed by app)

Scripts that upload here:

• src/deploy_to_hf.py → uploads deployment files

---

4. GitHub Repository

Repo URL: https://github.com/ananttripathi/engine-predictive-maintenance
Created by: You (manually via git push)

Files Uploaded:

Everything in the mlops/ folder, including:

• ✅ data/ (including engine_data.csv, processed/train.csv, processed/test.csv)
• ✅ src/ (all Python scripts)
• ✅ notebooks/ (EDA notebooks, etc.)
• ✅ .github/workflows/pipeline.yml ← GitHub Actions workflow
• ✅ requirements.txt
• ✅ Dockerfile
• ✅ README.md
• ✅ models/ (if you want to track model versions in git)
• ✅ mlruns/ (MLflow tracking data - optional)
• ✅ All other project files

How to upload:

cd /Users/ananttripathi/Desktop/mlops
git init
git add .
git commit -m "Initial commit: Predictive maintenance MLOps pipeline"
git remote add origin https://github.com/ananttripathi/engine-predictive-maintenance.git
git push -u origin main

---

Upload Workflow Summary

Step-by-Step Upload Process:

1. Data Registration → Hugging Face Dataset Repo

   python src/data_register.py
   

   • Uploads: data/engine_data.csv → HF Dataset Repo

2. Data Preparation → Hugging Face Dataset Repo

   python src/data_prep.py
   

   • Uploads: data/processed/train.csv and test.csv → HF Dataset Repo

3. Model Training → Hugging Face Model Repo

   python src/train.py
   

   • Uploads: models/best_model.joblib → HF Model Repo

4. Deploy App → Hugging Face Space

   python src/deploy_to_hf.py
   

   • Uploads: src/app.py, Dockerfile, requirements.txt, etc. → HF Space

5. Push to GitHub → GitHub Repository

   git add .
   git commit -m "Complete MLOps pipeline"
   git push origin main
   

   • Uploads: Everything → GitHub Repo

---

What Gets Uploaded Automatically vs Manually

Automatic (via Scripts):

• ✅ Hugging Face Dataset Repo → src/data_register.py and src/data_prep.py
• ✅ Hugging Face Model Repo → src/train.py
• ✅ Hugging Face Space → src/deploy_to_hf.py
• ✅ GitHub Actions → Runs automatically when you push to GitHub

Manual:

• ⚠️ GitHub Repository → You need to run git push yourself

---

File Size Considerations

Large Files (may be ignored):

• data/engine_data.csv → Uploaded to HF Dataset, but you might want to add to .gitignore for GitHub
• mlruns/ → MLflow tracking data (can be large) - ignored by HF Space deploy
• models/best_model.joblib → Uploaded to HF Model Repo, but you might want to add to .gitignore for GitHub

Recommended .gitignore:

# Large data files
data/*.csv
data/processed/*.csv

# MLflow tracking
mlruns/

# Model files (already in HF Model Repo)
models/*.joblib

# Python cache
__pycache__/
*.pyc
.venv/

---

Quick Reference Table

File/Folder	HF Dataset	HF Model	HF Space	GitHub
data/engine_data.csv	✅	❌	❌	⚠️ Optional
data/processed/train.csv	✅	❌	❌	⚠️ Optional
data/processed/test.csv	✅	❌	❌	⚠️ Optional
models/best_model.joblib	❌	✅	❌	⚠️ Optional
src/app.py	❌	❌	✅	✅
src/train.py	❌	❌	❌	✅
src/data_prep.py	❌	❌	❌	✅
Dockerfile	❌	❌	✅	✅
requirements.txt	❌	❌	✅	✅
.github/workflows/pipeline.yml	❌	❌	❌	✅
README.md	❌	❌	✅	✅

Legend:

• ✅ = Uploaded automatically or should be uploaded
• ❌ = Not uploaded to this location
• ⚠️ Optional = Can be uploaded but might want to exclude from GitHub due to size

---

Need Help?

• Hugging Face Dataset: Check src/hf_data_utils.py
• Hugging Face Model: Check src/hf_model_utils.py
• Hugging Face Space: Check src/deploy_to_hf.py
• GitHub: Standard git commands