Strategy for picking EC2 instance type for process with varying memory usage

Question

I have a process that I am currently running on EC2.

This is part of my web application, and I am relying on Elastic Load Balancer to automatically spin up more instances