Model deploymnet for real time stream

I have created a deep learning model using pytorch and I need to used real time streaming and batch prediction in evaluation, How could I deploy the model to accept data and evaluate at the same time? What are the tools that are used to do so LOCALLY on the same device?