MSc Data Science student in Dortmund. Working on automotive safety simulation and statistical forecasting — most of what's here is real project work, not tutorials. Currently restudying deep learning fundamentals toward a planned thesis on Bayesian Neural Networks for uncertainty-aware driver monitoring.
Published first-author paper on AEB failure modes — 100,000-scenario Monte Carlo simulation in CARLA, validated across 1.54M German crash records. → Read the paper
Built a time series pipeline (ARIMA / SARIMA / Holt-Winters) that matched Germany's official VDA registration forecast using only public data — backtested across 76 expanding-window origins.
5-horizon bankruptcy prediction pipeline (Random Forest, XGBoost, SMOTE) — AUC 0.903, built around a €28.5M cost-of-error argument.
Core stack:
Also comfortable with: Java, C/C++, React/Node (full-stack internship background), SAP S/4HANA, Power BI