Is there a way to run GPU accelerated ML inference on customer uploads?

Technical Q&A
1

Say a customer uploads a photo of their wedding ceremony. I want to be able to run my Tensorflow compiled model on that to suggest an add on based off the uploaded image.