Mustafa Sayed.

The problem

Reading a Quran app usually means a connection, an account and ads. The goal was a complete companion that works the moment it is installed, including where no usable network is available.

Approach

• ▸The full mushaf (114 surahs, 6,236 verses) ships inside the app bundle, with tafsir, translations and bookmarks stored locally.
• ▸Recitations from more than 240 reciters stream on demand or download for full offline playback, with a per-file download manager.
• ▸Prayer times are computed on-device, alongside a Qibla compass, an adhkar counter and a worship tracker.
• ▸Feature-first architecture with BLoC, Drift, GoRouter, GetIt and freezed, gated by a CI workflow on every push.

Outcome

• Works fully offline: no accounts, no ads, no tracking.
• Arabic-first interface with a night mode and a full right-to-left layout.
• Drift-backed local storage for the text, reciters, bookmarks and worship tracking.
• GitHub Actions CI plus a separate pipeline for the companion web documentation.

The problem

Fabric retail requires a storefront that feels tactile and premium, with quality that can be demonstrated rather than promised.

Approach

• ▸A DESIGN.md system as the single source of truth for every screen's look and feel.
• ▸Clean Architecture with a tactile, textile-inspired design language.
• ▸GitHub Actions CI plus an automated Android release pipeline.

Outcome

• 1,101 tests passing with 74.2% coverage, enforced on every push.
• Automatically signed Android releases produced by CI.
• A design system usable by both AI agents and human collaborators.

The problem

Tracking workouts, meals, sleep, water and BMI usually means five separate apps that never talk to each other.

Approach

• ▸Gemini-powered generators that build personalized workout and meal plans.
• ▸Firebase Auth + Firestore for accounts and user data.
• ▸Goal-based onboarding feeding a unified home dashboard.

Outcome

• Full tracker suite: workouts, meals, sleep, water, BMI, medicine reminders and photo progress.
• AI plans generated from the user's profile and goals.
• Cross-platform from a single Flutter codebase.

The problem

Most portfolio e-commerce demos depend on a backend. The goal was a production-grade architecture with no backend at all.

Approach

• ▸One Flutter app containing both storefront and admin, split only by role-based navigation behind a local PIN gate.
• ▸Shared domain and data layers across both roles.
• ▸Type-safe reactive persistence with Drift and a Result<T> error boundary.

Outcome

• 100% offline: local SQLite only, no Firebase, no Supabase.
• Sealed-class state handling with a strict error boundary.
• CI workflow guarding every push.
• 631 tests, all green: every layer, including end-to-end widget flows.

The problem

Flutter AI apps routinely embed API keys client-side, where they can be extracted from the built binary in minutes.

Approach

• ▸All AI calls routed through a backend proxy; the app never stores a key.
• ▸The security model is documented in the README as a public reference.
• ▸Gemini-style chat UI: message bubbles, reversed list, accessible labels.

Outcome

• Splash → onboarding → chat flow with light/dark themes.
• A documented, reusable pattern for safe AI in mobile apps.
• 17 passing tests across the onboarding and chat flows.