aroslaw Kutylowski speaks German, Polish and English . With DeepL, his startup's AI-powered translation tool, he can read and write in about 30 more.
That’s a tiny fraction of the 130 languages that you can translate using Google. But DeepL’s translations are calibrated for nuance by human editors and native language speakers. The startup employs 20 in-house editors and more than a thousand contract-based human translators and native speakers across the globe to assess the quality of translations produced by DeepL’s model and adjust them for accuracy.
In January 2023, the Cologne, Germany-based startup raised about $100 million in funding from global VC firms including Institutional Venture Partners , Atomico and Bessemer Venture Partners at a $1 billion valuation, according toDeepL has been downloaded on 25 million devices —a pittance compared to the more thaninstalls of Google Translate. But those who’ve used it laud its accuracy.
A Ph.D. in computer science, Kutylowski was developing the machine learning technology underlying DeepL in 2016, at the same time that Google researchers were working on Transformers, the translation-based technological breakthrough that powers ChatGPT. But as of now, DeepL does not use Transformers nor large language models for translation, he said, declining to specify DeepL’s exact architecture. “We knew that neural networks were going to dominate this field very soon,” Kutylowski said.