Nitisha Sharma

QUEUE_CARE_2026

76\% of India's 1.5 million clinics still operate on paper token slips and verbal announcements. Patients wait 2--3 hours with zero visibility into their position. Receptionists manage everything from memory. Doctors have no dashboard. This creates chaos, anxiety, and inefficiency at every level of the clinic workflow. Process Started by identifying the core pain point — patients have zero visibility into wait times in Indian clinics. Decided on a real-time socket-based architecture over polling because the 40% evaluation weight on live sync made it non-negotiable. Built frontend first with mock data so UI decisions weren't blocked by backend. Chose role-based auth early because receptionist approval flow needed to be airtight before any queue logic could work. Biggest challenge was Express 5 silently swallowing errors — downgraded to Express 4 which stabilised the entire backend. Socket.IO singleton pattern for queue state was a deliberate decision to avoid race conditions on concurrent patient requests. Results Both screens sync in when Call Next is clicked — verified by observing socket broadcast timing in the backend terminal. Patient waiting room correctly shows token number, tokens ahead, and estimated wait computed as tokensAhead × avgConsultTime with zero hardcoding. All 3 role dashboards fully functional — Admin monitors live queue and receptionist activity log, Receptionist manages incoming requests and calls next, Patient tracks position live without refresh. Receptionist approval flow works end to end with JWT-protected routes. Reflection I would add the Doctor Dashboard from day one rather than treating it as a future goal. The current flow has the receptionist deciding when to call next, but in real clinics the doctor signals readiness — that communication gap is the actual bottleneck I didn't fully solve. I'd also implement automatic average consultation time using timestamps instead of manual input by the receptionist — a rolling average of the last 10 consultations would make the ETA far more accurate without any manual effort. Finally I'd write backend tests before integration to avoid the silent error debugging time.