Simplest ML web service architecture

  amazon-web-services, docker, kubernetes, python, rest

In my current role, I work in data science; some DS build automation tools that "live" on a server somewhere. Other DS, like myself, work in decision optimization. The final output isn’t a living, breathing product but a decision recommendation for human action. In my new role, I need to deploy a model to web server. It should be able to receive excel files via an API push request, act upon the data, and return a result. There shouldn’t be a frontend whatsoever, as the target audience will be data analysts who use the excel push request, likely from a Jupyter notebook, using python.

In this question, I’m requesting a very broad description of the simplest option(s) available to me. For example, I’m not sure AWS/Docker would make things streamlined or add additional complexity that I don’t need.

Any additional thoughts are welcomed.

Source: Docker Questions

LEAVE A COMMENT