I’m working on an open-source python tool to spin up Deep learning EC2 machines in a single command. The goal is to make use of EC2 machines for development without fiddling with the Console, managing SSH keys.
tldr: This tool reduces all the steps in AWS EC2 Setup to a single command.
Here is a quick 90-sec demo:
GitHub repository:
It takes less than a minute to set up and try it. I’ll hang around on this thread to listen to your feedback.
The tool is very simple and sets up a secure EC2 machine that is fully configurable. It quickly gets out of the way and lets you configure it the way you want it. It retains all your packages, code and data after every session and lets you resume when you need it.
The tool runs completely on your machine - there are no servers. So you only pay for the AWS machines and storage you use. You can use your AWS student credits as well.
One last thing! The tool also lets you figure out what is the EC2 quota limit for the instance type you want to use. It also lets you easily request a quota increase from the command line!
Good question! The goal of this tool is to let you develop ML solutions on EC2 with minimal work.
The development machines are highly configurable - you can choose any instance type for working on Jupyter notebooks (eg: t3.large) and then quickly switch to a powerful gpu machine (p3.xlarge) when you need the computation power for training.
While other solutions like Salamander are full web applications, infinity is a just a cli tool you install on your laptop. It uses vanilla EC2 apis to create, start and stop these machines. That means you can easily use your existing EC2 configuration for infinity.