Serving ML model in production. Tensorflow-serving and autoscaling

Question

I\'ve been serving an image classification ML model over HTTP.
Running Web server and model as single container unit.
I did not use tensorflow-serving.
Archit