JotMe — Real-Time Multilingual AI Platform
A multilingual AI platform for real-time translation, transcription, meeting intelligence, and cross-language communication across online meetings, live events, and in-person conversations.
I have taken major ownership across the JotMe product ecosystem — technical design, frontend and backend systems, real-time communication, AI integrations, desktop and mobile applications, production deployment, debugging, and long-term product development. Responsibilities extend well beyond isolated feature implementation into architectural decisions, re-architecting existing systems as requirements evolve, and investigating production issues across the stack.
Move audio through capture, transport, session routing, and multiple AI providers fast enough to feel real-time — while surviving disconnects, provider failures, cross-platform audio quirks, and multiple concurrent outputs (translated speech, source transcript, translated transcript, UI events).
Chose persistent WebSocket sessions over request/response for continuous bidirectional data, separated browser notifications from page-scoped chat rendering into a global authenticated provider, ran independent processing pipelines for outputs with different latency and correctness requirements, and abstracted shared product logic from environment-specific desktop, mobile, and browser code.
Built and evolved audio-processing pipelines for multilingual speech translation and transcription — streaming audio, persistent WebSocket sessions, session lifecycle management, transcript synchronization, low-latency delivery, and reconnect/error recovery.
Electron desktop app used to bring real-time translation into online meetings. Handles physical and virtual audio routing, microphone and system audio capture, device management, cross-platform OS differences, installer workflows, and real-time backend communication.
Multilingual transcript processing and presentation — streaming partial and final events, speaker separation, mixed-language content, transcript grouping, non-duplicating UI updates, and synchronized transcript state across clients.
Re-architected browser notifications from a page-scoped implementation into a global authenticated notification system, so users receive new-message notifications across the app without duplicates inside the chat experience. Covers real-time messages, thread replies, unread state, email + browser notifications, background/unfocused tab behavior, and tab title / favicon unread indicators.
Authentication, account management, subscription systems, billing-related flows, settings, referral systems, permissions, meeting history, product configuration, and internal dashboards.
React Native application for live in-person interpretation, mobile audio workflows, real-time translation and transcription, sharing backend systems with web and desktop for cross-platform product consistency.
- Enabled multilingual real-time communication across web, desktop, and mobile
- Extended product functionality beyond a single application platform
- Improved notification behavior across the entire authenticated application
- Reduced duplicate and inconsistent real-time behavior
- Supported production systems used by real customers
- Built internal tools needed to operate and support the product

