Really great write up, a lot of moving parts to get right when you want to deploy an ML model to production. Docker and FastAPI are great tools, but they still require a bit of configuration for real production deployment and scaling.
We've created a platform that helps abstract away the API creation and deployment complexity, and allows you to simply upload a project, define your entry-point function and get an instant scalable API service that is shareable AND monetizable.
You can check it out @ www.byvalue.org, or check out our tutorial here https://medium.com/byvalue/how-to-create-your-very-own-monetized-api-79546a81c81d?source=friends_link&sk=b5817af292b60d7d8ad7f16bfffa3afe
Would love to connect & hear your thoughts about the platform!