AWS EC2 cost optimization solution

I have Java app that do complex media processing and gives result after X hours (anywhere from 2-12 hours).

What I am looking for a solution that would take

a) Job specific parameters inputs like how much memory/RAM/Time Java application need and based on input it will create a new instance of EC2.

b) It should use available instances if they are free or stopped, or launch new ones.

c) Cost optimization is a MUST.

d) Just want to focus on business logic instead of dealing with task/worker instance going down, broken network connections, etc.

NOTE: Have one instance of EC2 where job manager will be running and based on request received new instance of EC2 will be created.

Thanks