POST/api/v1/projects/medreseai202 ACCEPTED · IN DEVELOPMENT

MedreseAI

Duolingo mechanics meet religious education — with AI listening to your pronunciation.

// hero capture pending — placeholder rendered in-house
6
learning levels
14
edge functions
STT
AI pronunciation scoring
offline
first architecture

#01 The product

MedreseAI is a gamified Islamic education and ibadah utility app: structured learning paths across six levels with Duolingo-style mechanics — XP, streaks, badges, daily missions, and leaderboards — combined with daily worship tools.

The standout feature is AI pronunciation assessment: users recite, speech-to-text transcribes, and a similarity-scoring pipeline grades pronunciation. A placement test routes learners to the right level, and a special-days engine adapts content to the religious calendar.

Utility features make it a daily-use app rather than a study app: prayer times, Qibla finder, Ramadan mode, prayer and habit tracking, dhikr counter, and an Ayah/Hadith/Dua library with bookmarks.

#02 Engineering

Expo SDK 54 + React Native with TypeScript strict mode and NativeWind for styling; TanStack Query and Zod for data flow. The backend is Supabase end-to-end: Postgres with row-level security, Auth, Storage, and fourteen Edge Functions covering attempt submission, mission lifecycle, pronunciation assessment, prayer times, and content embedding.

Semantic search and an AI tutor with citations run on a content-embeddings pipeline over the library. Game economy integrity is enforced server-side — XP, streaks, and badges can only be written by Edge Functions, never the client. The app is offline-first with home-screen widget scaffolds for iOS and Android.

#03 Status

In active development under the DevJoys studio — schema, edge functions, gamification loop, and learning engine are built; content pipeline and store release are in progress.

Want the full story behind this build?

POST /contact