Portfolio Project: Demonstrating AI/ML application to real-world telecom challenges using domain expertise from 10+ years in network operations.
| # | Project | ML Type | Algorithm | Target | Result |
|---|---|---|---|---|---|
| 1 | Churn Prediction | Binary Classification | XGBoost | is_churned |
AUROC: 0.86 |
| 2 | Root Cause Analysis | Multi-class Classification | XGBoost | is_root_cause |
Acc@1: 0.91 |
| 3 | Anomaly Detection | Unsupervised | Isolation Forest | label_anomaly |
F1: 0.70 |
| 4 | QoE Prediction | Regression | LightGBM | mos_score |
RMSE: 0.45 |
| 5 | Capacity Forecasting | Time-Series | LightGBM+Prophet | traffic_load_gb |
MAPE: 14.5% |
| 6 | Network Optimization | Reinforcement Learning | Q-Learning | KPI improvement | +61% vs random |
- Domain Expertise: Deep understanding of telecom network operations and challenges
- Problem Framing: Translating business problems into well-defined ML tasks
- Data Engineering: Hand-crafted synthetic data generators with embedded telecom physics
- End-to-End Thinking: Data generation, feature engineering, modeling, evaluation, and business insights
- Communication: Clear documentation for both technical and business audiences
| Category | Tools |
|---|---|
| Language | Python 3.11+ |
| Package Manager | uv |
| ML Frameworks | XGBoost, LightGBM, scikit-learn |
| Data Processing | Pandas, NumPy |
| Visualization | Matplotlib, Seaborn |
| Interpretability | SHAP |
| Testing | pytest |
| Linting | Ruff |
| CI/CD | GitHub Actions |
Each project is fully independent. To explore one:
cd 01-churn-prediction
uv sync
uv run python -m churn_prediction.data_generator
uv run jupyter lab notebooks/See each project's QUICKSTART.md for detailed instructions.
Every project follows the same structure:
0X-project-name/
├── .github/workflows/ci.yml # CI pipeline
├── data/
│ ├── raw/ # Generated synthetic data
│ └── processed/ # Feature-engineered datasets
├── src/{package_name}/
│ ├── __init__.py # Package exports
│ ├── config.py # Configuration management
│ ├── data_generator.py # Domain-informed data generation
│ ├── features.py # Feature engineering pipeline
│ └── models.py # ML model implementations
├── notebooks/
│ └── 0X_analysis.ipynb # Main analysis notebook
├── tests/
│ └── test_data_quality.py # Data quality tests
├── .gitignore
├── pyproject.toml
├── README.md
├── QUICKSTART.md
└── CONTRIBUTING.md
Adityo Nugroho Telecom Professional | AI/ML Practitioner
All projects are MIT licensed for educational and portfolio purposes.