I specialize in Forecasting, Prediction Modelling, and Reinforcement Learning.
I love turning complex data into actionable insights—whether that's optimizing retail forecasts or training AI agents to beat me at board games.
🤖 Machi Koro Reinforcement Learning Agent
An AI agent trained to master the city-building card game Machi Koro.
- Tech: Python, Stable-Baselines3, PPO Algorithm, Self-Play.
- Highlights: Utilizes a custom Gymnasium environment where the agent learns to beat random and heuristic-based opponents via self-play curriculum learning.
⚽ Master Thesis: Sports Analytics & Home Advantage
Winning more without improving skills: The impact of familiarity during Covid-19.
- Tech: Python, Web Scraping, Statistical Modelling.
- Highlights: Analyzed 1,785 matches across 6 European leagues to determine if "Home Advantage" still exists without fans. Discovered significant correlations between field surface familiarity (artificial vs. grass) and winning probabilities.
🧠 Master Thesis: Coach Interventions in Soccer
Never change a winning team? Assessing coach impact using ML.
- Tech: Random Forest, MLP (Multi-Layer Perceptron), Linear Regression.
- Highlights: Modeled minute-by-minute match data from the "Big Four" leagues (18/19 - 21/22). Found that increasing substitutions correlates with a decrease in final goals, and that MLP models outperformed traditional regression by 1.42%.
📚 Academic Citation Predictor
Predicting the future impact of scientific papers.
- Tech: NLP (TF-IDF), Random Forest Regressor, LazyPredict.
- Highlights: A machine learning pipeline that predicts the citation count of a paper based on its abstract, title, and metadata. Includes custom feature engineering for text analysis and log-transformation for outlier handling.
