They have a GitHub api hack that could work, but once I put content on the S3, it will basically do it for my automagically at the edge
Discussion
This is the way.
There is a next level of scale after that, but for now get the server out of the read path. Not just for cost but for sanity of operating with such a small team.
The write path you can switch to APIGW and a bit of lambda and eliminate the “running computers” thing there. Use cost savings plans.