Alexandre Rosseto Lemos

Implemented fraud detection models using XGBoost to minimize financial losses in real-time payments, saving over $500k/year for the company. Developed risk rules and data analysis methods in Python to stop fraudulent transactions in real time, saving over $300k/year. Performed portfolio management using data analysis and visualization, creating reports and monitoring tools to present portfolio results regularly to stakeholders. Built machine learning models to assist the Risk Operations team in prioritizing analysis and early fraud pattern detection, increasing overall fraud detection by over 50%. Developed a loss forecast model using monotonic regression to forecast losses for the financial department. Created customer segmentation models using K-Means clustering and psychographics to understand customer needs and preferences, providing insights for Trustly and its customers. Applied AB test experiments on different occasions using statistical tests to evaluate intervention impacts.

Implemented machine learning models using Python and scikit-learn with algorithms such as SVM, Random Forest, and MLP for fault detection in steel coils. Used SHAP analysis to understand root causes of faults. Developed an optimization algorithm improving performance over 15% using statistical analysis. Built machine learning models leveraging data analysis and Random Forest algorithms to predict steel temperature during processes.

Developed Data Warehouse projects using dimensional modeling to feed dashboards and reports for multiple clients. Created ETL pipelines using Pentaho to extract, transform, and load data from various sources. Built dashboards and reports in Tableau and Power BI to track KPIs and provide self-service tools for stakeholders.