At Raiffeisen Bank International, I built an XGBoost model on 300M+ card transactions to predict which new cardholders keep using their cards after the first transaction. SHAP driver analysis informed changes to the bank’s activation campaigns, and I designed A/B tests to compare communication and bonus strategies.
I also clustered Eurostat SME survey data into personas adopted as target groups for the bank’s SME project. I automated a CRM KPI framework of 12 management-level metrics used in quarterly and annual budget planning.
At A.P.I. Churi Distribution LTD, I automated a pricing pipeline that extracted supplier data from unstructured emails into a custom BI tool, enabling real-time vendor comparisons and identifying potential annual COGS savings of $100K+. Statistical forecasting and machine learning helped cut label inventory waste by 20%, and I integrated fragmented data sources into an internal data product used daily by 30+ employees.
At WU Vienna, I benchmarked seven machine-learning models against the Beneish M-Score for financial statement fraud detection; the best model outperformed the benchmark at every detection threshold on a five-year held-out test. My thesis examined Bitcoin and AI-equity tail risk, and an industry capstone with IBM Austria resulted in a Streamlit text-to-SQL analytics app for non-technical stakeholders.
