I have a workflow within an application that includes image processing in various ways (resize, artifact reduction, region of interest, facial placement, person recognition). I have been doing this using a queueing system previously and some custom workflow, the process is actually rather complicated in order to make it cost effective. I have a new solution using Lambda instead of my previous process. So far it's working quite well.
It sounds like you've got quite a complicated process there, how did you find handling dependencies with Lambda? The docs include how to add npm dependencies but I couldn't find anything covering dependencies that the node modules might require. Does Lambda just magically figure it out?