World Cup 2026 Pool Optimizer
Elo ratings, Poisson goal models, Monte Carlo simulation, and simulated annealing applied to optimise a World Cup 2026 prediction pool bracket.
About
Medior/Senior Data Scientist (MSc Systems & Control, TU Delft) specialising in mathematical, statistical and algorithmic approaches to real-world problems. I like to design and productionise models and pipelines that scale — whether the domain is transport, energy networks, disease modelling, quantum error correction, or other complex systems.
I enjoy taking noisy, high-cardinality data and turning it into robust models and interpretable decision tools: model design, uncertainty quantification, efficient vectorised pipelines, and reproducible deployment. Always open to new challenges and puzzles to solve.
Stack
Data & ML
Python
Pandas
DuckDB
Ibis
Scikit-learn
Parquet
Databricks
LLMs
Agentic AI
GeoParquet
Infrastructure
Linux
Docker
Git
Kafka
Drone CI
Geospatial
QGIS
GTFS
OpenStreetMap
Valhalla
Learning
RAG
Projects
Overton Window Analysis
Using 29,591 parliamentary motions, SVD spatial analysis, and 2D extremity scoring to test whether the PVV election shifted the Dutch Overton window.
StemAtlas
Political compass derived from 25,000+ Dutch parliamentary votes using SVD embeddings and fused similarity search.
30km/h Speed Limit Study
GPS → map-matching → policy recommendation that led to measurable cost savings and operational improvements.
National PT Data Pipeline
Process 350MB+/day of KV6 data, optimized Parquet for scale.
Population Accessibility Analysis
Identified accessibility patterns for hospitals, GPs, supermarkets, and schools across the full synthetic population of the Netherlands — supporting national infrastructure policy decisions.
RAG Application
Knowledge management system to support monitoring and evaluation workflows.
Contact
Email: geboers.sven@gmail.com
Phone: +31 657 95 90 96 — The Hague area, Netherlands