Add ML Pipeline with FastAPI API Endpoints and Weights & Biases Integration

[p1rth13]

🚀 What’s New

This PR adds a complete machine learning pipeline for the Iris dataset with the following production-ready features:

• 🧠 Model training using Scikit-learn
• 🛠️ ML pipeline structure following best practices
• 🎯 Experiment tracking via Weights & Biases (W&B)
• 📊 Automatic confusion matrix visualization logged to W&B
• 🌐 FastAPI endpoints (/predict) for real-time inference
• 📂 Proper folder structure for models/, api/, and pipeline.py
• 📝 Updated .gitignore, LICENSE, README.md, and requirements.txt

---

📦 Files Added/Modified

• pipeline.py — Model training, evaluation, and W&B integration
• api/main.py — FastAPI app with /predict endpoint
• .gitignore — Added W&B logs, Python envs
• requirements.txt — Includes wandb, fastapi, uvicorn, scikit-learn, matplotlib, seaborn
• README.md — Setup, usage, and cURL test instructions
• LICENSE — MIT License
• PULL_REQUEST_TEMPLATE.md — This file

---

🧪 How to Test

1. Create virtual environment:

   python -m venv venv
   source venv/bin/activate  # or venv\Scripts\activate on Windows

2. Install dependencies:

   pip install -r requirements.txt

3. Run the pipeline:

   python pipeline.py

4. Start the FastAPI app:

   uvicorn api.main:app --reload

5. Test the prediction endpoint:

   curl -X POST "http://127.0.0.1:8000/predict" \
        -H "Content-Type: application/json" \
        -d '{"features": [5.1, 3.5, 1.4, 0.2]}'

---

📌 Notes

• W&B API key must be set via environment or CLI login
• FastAPI runs on localhost:8000 by default
• Confusion matrix visual is auto-logged to W&B dashboard

---

🧭 Related Issues or PRs

Closes #1 — Initial ML pipeline setup and deployment interface