Coding the interface would be the easy part.

But then: the content, the discovery, the algos, handling video at different resolutions, etc etc...