Founder of Privera, a Windows app that finds and removes personal data from PDF and Word files – entirely on your own computer. I'm studying industrial engineering (civilingenjör, industriell ekonomi) at Mid Sweden University.
- Privera – local redaction for Swedish documents. OCR for scanned files, text removed from the file instead of covered, and an audit log for every export. The source is private; the product is at privera.se.
- Rosetta – names the columns in BioLogic EC-Lab (.mpr) battery files that parsers don't recognise, using physics instead of a language model. 522 of 586 columns right on 7 public datasets it had never seen. On PyPI.
- Turncoat – measures how often tool-using AI agents get hijacked by prompt injection, with confidence intervals. Built for the Nebius × NVIDIA Global AI Hackathon. Live demo
- File Converter – Casio and TI calculator files to text and back, byte-exact, in the browser. Open it
Bachelor's thesis: automated risk analysis of all 5.4 million supplier transactions published by the City of Gothenburg (2021–2024), verified against public records. Read it in DiVA
Found and reported a bug in the galvani parser for EC-Lab files (issue #11), fixed in version 0.6.0.