Tsion Habtesilassei

I've built end-to-end machine learning systems for e-commerce fraud detection, BNPL credit risk modelling, and insurance risk analytics, combining data pipelines, feature engineering, model deployment, and explainability.

In my fraud detection pipeline with 10 Academy and Adey Innovations, I identified time-since-signup as the dominant fraud signal, addressed class imbalance with SMOTE, and delivered SHAP-based explainability. For Bati Bank, I engineered a proxy default label using RFM analysis and K-Means clustering, then shipped scikit-learn pipelines, MLflow tracking, FastAPI serving, and Docker.

At Oclock Motors, I delivered end-to-end IT support and infrastructure management to keep systems secure, reliable, and high-performing. At Dire Dawa University, I support computer engineering laboratories through system maintenance, troubleshooting, and hands-on practical instruction.

I'm currently deepening my MLOps and Generative AI knowledge through the Kifiya AI Mastery Program, bringing modular architecture, reproducible pipelines, rigorous testing, and CI/CD to data-driven work.