Projects
PAUL and the paulify CLI
I couldn't find an open source license that permitted personal and academic use while protecting against commercial misuse—so I wrote my own. While I was at it, I baked in some AI ethics requirements that work to protect marginalized peoples worldwide.
Thinking others might want something similar, I created a lightweight Python CLI tool for reproducible attribution under PAUL.
License Plate OCR: Russian versus U.S. Plates
Can rule-based vision pipelines reliably decode the chaos of real-world plates? This project for my Computer Vison course at CSU Global put that question to the test.
Over a two-week sprint, I developed an end-to-end pipeline using OpenCV and Tesseract, achieving a 50% character recognition baseline on Russian plates. I addressed plate skew and variable lighting with adaptive thresholds and transformations, gaining hands-on experience tuning Tesseract parameters and image morphology techniques.