I'm guessing it will take a bit of work to support multiple languages/frameworks/platforms. A simpler solution may be to allow an interviewer to spin up a machine with the image/environment of choice (AWS, DigitalOcean etc) and do some kind of screen sharing with markup - easy to circle some code on a touch screen tablet. Interviewers typically specialize in languages and platforms, so every interviewer can maintain their favorite "interview machine image".

The feature I like best is the group comments - I would recommend restricting read access by all interviewers until everyone's feedback has been read by the interview co-ordinator, and the everyone gets to see all the comments during the hire/not group discussion among a possibly distributed team.

Also, I'm wondering why take on (what I assume is) the rather onerous task of supporting voice chat? Wouldn't it be simpler to adopt Skype (thru an API?) rather than fight it? My humble two cents: focus on where you're adding value, not call quality, security etc.