I am working on Daestro[0], which is a cloud agnostic job orchestrator with built-in support for AWS, Vultr, DigitalOcean and Linode to run jobs on. Daestro can spawn and terminate compute instances based on requirement. It is suitable for running batch jobs or data engineering related jobs.

Self-hosted compute can also be linked to Daestro to run jobs on.

[0]: https://daestro.com