Hello 👋, I’m Eleftherios 👨🏻💻 a data scientist in Rotterdam 🇳🇱
I just finished my MSc Business Analytics & Management at RSM, Erasmus University Rotterdam, where I spent my thesis teaching a model to guess when delivery drivers ignore their planned route. Before that: customs data at the Port of Piraeus, CX design at Deloitte Digital, and a lot of basketball.
Latest projectA three-phase XGBoost framework that predicts when last-mile delivery drivers will deviate from their planned stop sequence, then proposes new sequences they are more likely to follow.
Forecasting how disruptive a scientific paper will be, using only what's known when it's published — with XGBoost and a Keras neural network on a 149k-paper citation database.
Replicating the Social Capital Atlas (Chetty et al., 2022, Nature): how cross-class friendships shape economic mobility in the US, rebuilt from open data in Python.
Repression and window dressing around Argentina's 1978 World Cup, replicated with regression models (Scharpf, Gläßel & Pearce, 2022, APSR).
An options-hedging engine for OpenBB Terminal: tell it a position, pick two options, and it solves for the weights that make the portfolio neutral to Delta, Gamma and Vega.
A multi-criteria evaluation of DALL-E, Midjourney, Stable Diffusion and Uizard for user interface design.
A social discovery app concept: personalised recommendations for places, experiences and events that best suit you and your friends.
An LLM pipeline that matches residents into small circles using embeddings, prototyped in 20+ hours and pitched to BCG.