Rovin Castelino

MemoryVerse AI '26

Queue Cure — Real-Time Clinic Queue Manager with <100ms Live Sync

76% of India's 1.5 million neighbourhood clinics use the manual method of managing patient queues through paper tokens and verbal announcements. Patients must wait for 2-3 hours with no visibility about their waiting place while the staff has to handle everything with memory. There is no feedback system; patients go away, clinics lose money, and the doctors don't get any dashboard. What is required: A simple software without installation that gives real-time control to the staff by clicking and data-driven waiting time to patients on their smartphones. Predefined three distinct roles at the beginning: receptionist, patient, display screen. Started with building database schema — singleton queue_settings table and patients table with RLS policies so that all the screens have sole source of truth with zero authentication cost. Selected Supabase Realtime over polling for sub-100ms WebSocket synchronization for the lobby screen. What went wrong: UI of the manual override feature relied on local state, leading to the situation when field disappeared without being saved in the database. Resolved by implementing UI solely based on database state. Final refactors were deleted 46 unused boilerplate components, reduced CSS size by 48%. Under 100ms live synchronization for all 3 screens. Under 2 seconds to sign up a patient. Removing 46 unnecessary elements saved 48% of CSS package size (96 kB to 50 kB). The waiting time was calculated based on actual consultation length, not on estimations. For next time: doctor’s screen with consultation timer, QR codes for signing up with a token, and daily analytics dashboard for clinic owners. I would begin with a proper concurrency stress test before beginning UI work, as the manual override problem would be found right away by testing in advance. I would include a screen for doctors that has a consultation timer and patient notes, thereby increasing the usefulness of the application outside of the waiting room. I would make sure to include QR codes at check-in so that users never need to recall their token number. Finally, I would create an analytics dashboard for the clinic owner that shows peak hours, average waiting times, and skip rates.