This project was motivated by my frustration of solving CAPTCHAs on the Bundesanzeiger and Unternehmensregister websites while performing financial analysis. To eliminate this bottleneck, I trained an OCR model capable of solving these CAPTCHAs in under 50 milliseconds.
I packaged the solution into a public Chrome extension, allowing anyone to bypass this repetitive verification step seamlessly. The result was a faster, smoother research workflow that removed a common pain point for analysts and researchers.
In addition, I prepared and delivered a presentation on this project as part of a Data Science course with my group, showcasing the OCR model design, training process, and deployment as a real-world application.