Delphin K.

Data Scientist | Fraud Analytics Engineer

Fraud Detection • Risk Analytics • Data Engineering

I use machine learning, SQL, and data pipelines to track risk patterns, optimize operations, and protect financial infrastructure.

My work sits at the intersection of machine learning classification, risk analytics, and modern data engineering. I bridge the gap between raw data and real-world risk mitigation, designing end‑to‑end systems from SQL pipelines and Airflow orchestration to high-precision fraud detection models and real-time monitoring tools that support fraud operations teams.

• Fraud Analytics & Systems: Real‑time scoring engines, transaction monitoring, and high‑precision machine learning models engineered to stop financial crime.
• Risk Modeling: Credit risk segmentation, anomaly detection, and operational data validation to maintain strong financial controls.
• Data Engineering: Building robust, scalable pipelines using Airflow, dbt, and SQL to process large-scale event data and maintain clean architectures.
• Operational Monitoring: Deploying FastAPI microservices, Dockerized pipelines, and live data tools that give risk operations teams real-time visibility.
• Analytics Strategy: Turning messy transaction signals into clear, actionable metrics and insights that help teams catch bad actors faster.

• 🛡️ Fraud Detection Engine: High-precision machine learning classification system achieving 0.93 precision and 0.82 recall on imbalanced financial transaction data.
• 📊 Metro Transit Analytics Platform: Built scalable data pipelines processing over 2.1 million daily API events using Apache Airflow and Docker.
• 📉 Credit & Portfolio Risk Analytics: Scorecard‑driven segmentation and predictive risk analysis to support commercial lending and decisioning.
• 🚲 Demand Forecasting & Retraining: Cloud‑deployed forecasting API with 51% MAE reduction using automated retraining data pipelines.