I build end-to-end data and AI systems from raw data ingestion and pipeline architecture to production ML models with real-time APIs. I don't just train models; I ship them. Every project I build goes from notebook to deployed, tested, containerised product.
- Building production ML systems — ensemble models, MLOps pipelines, and explainable AI
- Designing data pipelines — ingestion, transformation, feature engineering, and validation
- Deploying AI-powered APIs — FastAPI, Docker, real-time inference with SHAP audit trails
- Strong foundation in statistical analysis, class imbalance handling, and model evaluation
- Currently deepening expertise in large-scale data engineering and AI system design
- Background in blockchain development — bringing systems thinking from Web3 to AI
Production-grade fraud detection on 284,807 real credit card transactions.
- Stack: XGBoost + Random Forest stacking ensemble → Logistic Regression meta-learner → FastAPI → Docker
- Results: AUC-ROC 0.9755 | F1-Score 0.8555 | 74/95 fraud cases caught | 6 false alarms on 56,746 test transactions
- Pipeline: 4 notebooks → src/ production modules → REST API → containerised deployment → 24 automated tests
- Explainability: Real-time SHAP audit trail embedded in every API response
- Email: dammifabz@gmail.com
- Twitter: @damilola356075
- LinkedIn: Fabunmi Richard