There are a few ways to approach deleting unused resources, perhaps tagging items that are part of some trusted or known group, as some have suggested here and then manually monitoring for resources that isn’t tagged appropriately.
You could then extend this to a python script that you run every so often to ID and delete these resources.
However, there are other (potentially less intensive) options.
1) AWS already provide a way for us to monitor our resources, identify unused ones and terminate them with CloudWatch alarms. You can start by configuring some alarming based on some set of checks that you might already be doing manually. Once you are comfortable that the alarming is accurate, you can add a termination step to the chain.
2) Alternatively, our friends at Netflix made this chill tool called JanitorMonkey to monitor and delete unused Instances, EBS Volumes, EBS Snapshots and Autoscaling groups. It is part of the original so-called “Simian Army” open source pack from Netflix and written in Java. Even thought it’s a few years old now, this thing is highly configurable and well worth a look.
Both options here allow you to configure warnings and business rules that fit your needs. Additionally, using CloudWatch might also be a good opportunity to learn more about AWS logs and anomaly identification.