Multinode

Hello Product Hunt 👋 Multinode's alpha launches today! 🧑‍💻 Who are we? We're a small team of engineers that spent waay too much time setting up dynamic user-triggered workflows with Kubernetes when working on our previous product, even though conceptually we knew exactly what our system is supposed to do [1]. 🚀 The product Multinode helps you eliminate the hassle of setting up compute infrastructure so you can build arbitrarily complex compute without leaving the comfort of your local Python environment. You can use it to set up distributed async workflows, continuously running daemons, APIs, etc., with a help of only 5 additional annotations! 🔧 Technical details Internally all the infra is on AWS. All the compute runs on ECS with Fargate using controllers similar to those used by Kubernetes. Unfortunately, Fargate has pretty annoying resource limitations (hence https://multinode.dev/docs/resou...) so we will likely port it to ECS on EC2s or just straight to K8s at some point. The task registry and multinode's dict is running on Amazon Aurora. We use Python for the actual product and docs run on Firebase with next.js using one of the official Tailwind templates (I'd really recommend them!). [1]: For curious, it was something similar to our chess example: https://multinode.dev/docs/core-... PS: join the rest of the community on our discord: https://discord.gg/uSkprP9K

Nice team, congrats on the launch! It seems like a great tool for data/model guys.

looks like a nice abstraction for distributed computing :) where does it run?