I am somewhat in that class. Figuring out ways to horizontally scale video processing at the scale of YouTube sounds like a neat problem.

Obviously pretty much anyone here can get an extremely basic YouTube clone done in an afternoon or two. Spin up RabbitMQ, write an upload web server, transcode the video with ffmpeg and store it somewhere, serve it via HTTP. That’s trivial, but YouTube has to deal with 500 hours of new video every minute [1]. At those levels, the basic “senior engineer solutions” to problems stop being as appropriate, and I think those kinds of problems are ridiculously fascinating.

The annoying thing is that since YouTube has a monopoly and I have somehow managed to fail Google’s personality test multiple times, I don’t think I’ll ever get a chance to work on that kind of problem.

[1] https://www.globalmediainsight.com/blog/youtube-users-statis...