6 MB Tesseract (with English training data) to fit inside AWS Lambda - View it on GitHub