react-native-executorch

Why we included this project

On-device inference in React Native has usually meant fighting with native modules, and this library tries to smooth that path. It loads ExecuTorch-exported models and runs them entirely on the phone, covering object detection, segmentation, OCR, and text generation. Because models stay local, you avoid server round trips, which matters for privacy-sensitive content like documents or medical records and for apps that need to work offline. The repo includes prebuilt modules and example integrations, so you can test a model on real hardware before committing to a feature, and swapping models during prototyping is straightforward.