ZED (My Islamic Reminders)

Role: Contributing developer (team project, led by another engineer)Timeline: 2025-2026Type: Cross-platform app + admin

Expo (React Native)

Next.js

Convex

OpenAI TTS

TanStack Table

Recharts

ZED (My Islamic Reminders) hero image

ZED is a spiritual broadcast platform built as a monorepo with three parts that share one backend: an Expo mobile app for end users, a Next.js admin dashboard for the people running it, and a Convex backend that holds the data and does the scheduling, notifications, analytics, and audio generation. The product sends users short Islamic broadcasts at their preferred local time, in their chosen language, and automatically turns each one into narrated audio with word-by-word highlighting as it plays.

The hard part of this product isn't showing content. It's orchestration. Users want broadcasts delivered at a specific time in their own timezone, in one of several languages including right-to-left scripts. Each broadcast should be consumable as text, as audio, or as something shareable. Administrators need to see what's actually happening: what's been delivered, viewed, and played, and whether audio generation is keeping up. And when someone misses content, the system should follow up on its own. The real work is pulling scheduled delivery, language selection, audio generation, push notifications, and analytics into one coherent system.

This was a team project led primarily by another engineer, and I contributed as one of the developers on it. My involvement was focused rather than primary, within an architecture built around Convex as the single source of truth shared by both the mobile app and the admin dashboard. The design principle across the codebase is that the backend owns the orchestration, including scheduling, retries, audio jobs, and analytics, while the two front ends stay thin and read from the same generated API.

MobileExpo and React Native with Expo Router, Jotai, expo-audio, expo-notifications

AdminNext.js with React, TanStack Table, Recharts, Sonner, and a shadcn-style UI

BackendConvex for data, queries, mutations, storage, cron jobs, and a background work pool

MediaOpenAI text-to-speech for narrated audio generation

Notificationspush delivery with a retry policy

Personalised scheduled deliveryeach user's broadcast arrives at their preferred local time, matched to their timezone.

Multilingual contentbroadcasts in several languages, including full right-to-left support, with the best variant chosen from the user's preferences.

Auto-generated narrationevery broadcast is turned into audio through text-to-speech, with word timings so the text highlights in sync as it plays.

Guided broadcast experiencethe mobile screen plays narrated audio, highlights along, and lets users share the text.

Admin operations consoledashboards for users, delivery, views, plays, and audio-generation progress, plus growth and language-performance reporting.

Broadcast and content managementcreate and edit broadcasts, regenerate or retry stalled audio jobs, and manage localised app copy and store links.

User management and settingsmanage roles and active state, choose single or multi-reminder behaviour, and send test notifications.

ZED (My Islamic Reminders) screen 1

ZED (My Islamic Reminders) screen 2

Audio generation is the other moving part that has to stay reliable. Turning text into narrated audio is slow and can fail, so each piece of localised content tracks its own generation status and retry count, and admins can see progress and re-run stalled jobs from the dashboard. Engagement tracking, recording delivered, viewed, and played events with timestamps, is built directly into the data model, which means reporting comes straight from the same store without a separate analytics pipeline.

All projects Next projectSchooliee

Overview

TheProblem

MyRole&Approach

TechStack

KeyFeatures

Gallery

Challenges&Solutions

Outcome