Virtual Classroom Feature Spec Template

Almost always use a third-party SDK (Agora, Twilio, Daily, Vonage). Video infrastructure is enormously complex: codec negotiation, TURN servers, adaptive bitrate, echo cancellation. The differentiation in an EdTech virtual classroom is in the engagement and collaboration features built on top of the video layer, not the video layer itself. Budget 3-6 months for the engagement features, not for reinventing WebRTC.

For lectures: instructor video + shared content is sufficient. Learner video is optional and usually off in 50+ person sessions. For seminars and workshops (under 30 people): support up to 25 simultaneous video feeds. For study groups: up to 8. Do not try to display 100 video tiles. It creates performance problems and nobody can see anyone clearly at that scale.

Polling. It requires the least engineering effort and has the highest impact on instructor adoption. Instructors who can throw a quick poll into a lecture see immediate participation data. Build polls first, then hand-raising, then breakout rooms. Whiteboards are high-effort and lower priority unless your use case is specifically workshop-oriented.

Store all session times in UTC. Display times in the learner's local timezone with explicit timezone labels ("2:00 PM EST / 11:00 AM PST"). Send calendar invitations with proper timezone data so external calendars render correctly. For global audiences, offer multiple session times or record sessions for async viewing.

That depends on your pedagogy. If live interaction is the core value (workshops, seminars), make attendance meaningful by including participation scores and breakout room activities that cannot be replicated from a recording. If the content is primarily lecture-based, let recordings serve as the primary consumption mode and use live sessions for Q&A and discussion only.