Data scientists leverage cloud providers for embedded machine learning functionalities via SaaS services, transitioning from trial versions to costly, less customizable solutions. This article proposes an NLP project structure for ML, focusing on an OCR model integrated into translation systems. Utilizing parallel processing, Bert Transformers Models, IaaS like Docker, Python functions, and computer vision CNN, the model ensures real-time predictions.