I started working on this on March 15th, and the past few days was about testing everything, preparing the release and squashing bugs. The advantage of this type of functionality is that there are tons of high quality papers and articles already written on the subject, especially this one helped me a lot: http://www.codecommit.com/blog/java/understanding-and-applyi.... There were a lot of sleepless nights involved too.
I'm not sure about the VS Code and Atom teams, but I imagine that there are more constraints and there is more functionality to implement since they work with hundreds of potentially big files. I think those factors make it harder to implement this functionality than in the case of CodeSandbox.
Maybe March last year.
And if you really want to make sure, we're open source: https://github.com/CompuIves/codesandbox-client.
I'm impressed that you can build and ship this with a small team and short time, whereas the VS Code and Atom teams are still in beta with their live coding features. I understand scope and scale are very different, but this is a great accomplishment!