Learning AI Deployments

I've been learning how to deploy and augment LLM models using Fly.io's
GPUs. If you are building AI apps, here are five things I've learned
as I get more familiar with everything:
1. Every LLM model should be its own application
2. Don't allow the public to access your models.
3. Don't redownload your base models on every run.
4. Don't wipe your database.
5. Stop idle GPUs.
If you want to learn more about any of these topics, dig deeper a bit
into the tech, add a comment (or just say hi!)

[@julianduque]

Mr Builes! Good to see you