Featured Projects
Personal Grocery Inflation Index
Behavioral CPI modeling from real receipts (2023–2025). This research-style deep dive project reconstructs a personal inflation index from over 600 Albert Heijn receipts, parsed using my custom OCR pipeline. It combines classical CPI logic, behavioral weighting, and structured microdata to explore how inflation is actually experienced. Results are benchmarked against official CBS CPI figures for food and non-alcoholic drinks.
Albert Heijn Receipt Parser
A purpose-built OCR and text extraction pipeline to convert PDF grocery receipts into clean, structured datasets. Handles messy line items, bonus flags, and multi-user support for downstream CPI analysis.
Voice Deepfake Detector (ASVspoof2019)
Lightweight voice deepfake detector built in a one-day sprint. Uses Wav2Vec2 embeddings and logistic regression to classify synthetic speech. Clean, constrained, and sharply scoped. No tuning, no GPU, just working results.