This project was motivated by frustration with solving CAPTCHAs on Bundesanzeiger and Unternehmensregister websites during financial analysis. I trained an OCR model capable of solving these CAPTCHAs in under 50 milliseconds to eliminate this bottleneck.
The solution was packaged into a public Chrome extension for widespread use. This enables users to bypass repetitive verification steps seamlessly, creating faster research workflows and removing common pain points for analysts and researchers.
Additionally, a presentation on this project was prepared and delivered as part of a Data Science course with collaborators, covering OCR model design, training methodology, and real-world deployment applications. You can download the presentation.