Open-source IDE for data-centric NLP

refinery enables us to create high-quality training data within hours instead of weeks. You can tell that it is designed by and for Data Scientists, and we can use it for several use cases. I wouldn't want to miss it.

With refinery we were able to drastically simplify and organize a process of automated, standardized content classification, while simultaneously creating an implicit labeling documentation to rely on. The rich and versatile Python SDK allowed for a seamless integration with our pre-existing architecture.

I have worked very closely with Johannes and the Kern AI team and it was always struck by excellent expertise, deep analytical skills and customer orientation. I am sure Johannes and his team will have a great future.

We'll be able to further improve the automatic selection of conspicuous medical bills, which are routed to the final expert check at Barmenia.

The example of Kern AI shows how closely research and entrepreneurship are intertwined at HPI. The system, which emerged from a research project, helps data scientists to implement precise AI models significantly faster and to integrate business departments into the development process.