Pali is considered as the language of the canon continued to be influenced by analysts and grammarians, and by the native languages of the countries like Sri Lanka, in which Theravāda Buddhism became established over many years. The Pali language has been used to write many stone inscriptions in ancient times. For ordinary travelers the recognition of the content of ancient inscriptions and some other written material are not possible. This study has focused to find a solution with a mobile application to recognize Pali characters in a user-friendly User Interface. The character recognition in real time is the most essential part of the study. Machine learning and neural networks are trending technologies adapted for handwriting recognition in some languages. But for many languages this technology has not been developed yet. Pali is one of such languages that had survived until the eighteenth century. In this study the images of Pali letters are identified through a trained Convolution Neural Network (CNN). Python and Android Studio software were used for training process and identifying letters respectively with the developed mobile application. The limited capacity and the pro-cessing power of a mobile phone makes it more challenging to run the application. TensorFlow Lite is end to end open source platform in machine learning. Therefore, TensorFlow Lite was used in this study. Since the Android mobile is a common equipment which everybody has in modern societies, using this Off-line Pali character Recognition mobile application the archelogy researchers and travelers can use them conveniently to understand the content written in ancient documents.