如何在EC2上部署机器学习模型并连接到没有Lambda的API网关？

Question

I have trained a classifier using scikit.learn. I want to set up an API Gateway, so a user can submit the data using this API. After data run through the trained model, return the result using the API as well.

Since my model requires spacy.io's big trained model as the backend, I CANNOT use Lambda due to Lambda's limitation on file size (512 MB). But all I Googled is related to Lambda.

I am thinking, in my EC2 instance, I setup a Flask or Django app to receive the data from API Gateway and then run the model, return the result back to API Gateway.

But I don't know how to do it, could any one point out some resources?

Or is there any better solution out there?

Answer 1

It sounds like you've only discovered documentation for the Lambda integration method. You need to look at either the HTTP or HTTP_PROXY integration methods. I would start by reading this page in the documentation, and then proceed to this page .

If you also want to prevent anything but API gateway from directly accessing your back-end system, then you will also need to setup client-side SSL certificate verification on your back-end.