Muhammad Rayan
1 profile view
Summary
CS student at FAST-NUCES building real products for Pakistani markets. Experienced in Flutter, Python, and full-stack web with 4 shipped projects spanning mobile apps, AI/ML pipelines, and early-stage startups.
Skills
Certifications
Google UX Design Professional Certificate — Coursera / Google
Issued 2025-01-01 · No expiry
AWS Cloud Computing Workshop — FAST NUCES (6-Week Program)
Issued 2026-01-01 · No expiry
Projects
AroundFAST — Campus Accommodation Finder
Cross-platform mobile app that lets university students across Pakistan search, filter, and review nearby hostels. Used Supabase with PostgreSQL as a fully serverless backend. Handles real-time hostel listings and a live user review system with no self-managed infrastructure. Wrote a custom string-parsing filter for rent ranges since hostels list prices as free-text (e.g. '5k to 8k'). Dropped OpenStreetMap with async geocoding instead of Google Maps to keep it free. Every hostel gets a dynamic location pin with zero API cost.
FlutterDartSupabasePostgreSQLOpenStreetMapSAILAB — Flood Early Warning System
Built a flood monitoring app for KPK that pulls live river discharge data from Open-Meteo across 9 gauges and scores each against a 5-tier risk threshold to flag dangerous conditions in real time. Automated emergency alerts in English, Urdu, and Pashto via Twilio SMS and voice calls so district officials can reach village reps the moment a river breaches a safe level. Runs background checks every 15 minutes using WorkManager isolates and fires system-level push notifications on critical breaches, with no active user interaction needed. Role-based login via Supabase Auth with bcrypt and JWT. Shelter discovery uses a hybrid approach: pulls from the app’s own DB first, falls back to live OpenStreetMap Overpass API queries.
FlutterRiverpodSupabaseTwilioOpenStreetMapAura — AI-Powered File Organizer
Desktop app (Linux/Windows) that runs a 7-stage ML pipeline on any folder: K-Means clustering groups similar files, KNN scores their importance, A* suggests a clean folder structure, Isolation Forest catches anomalies, and Linear Regression predicts which files will become junk. Tracks user decisions (keep or delete) and uses them to retrain the underlying models over time, so the app gets more accurate the more you use it. Generates a full PDF audit report with cleanup ROI, security risks, and storage trends. Packaged with PyInstaller into a single executable so there’s nothing to install.
PythonPyQt5scikit-learnmatplotlibFake Financial Receipt Detection System
Detects forged financial receipts using three approaches in one pipeline: OpenCV for pixel-level tampering, a trained ML classifier on real vs. fake receipt data, and text extraction to flag numerical inconsistencies. Combining all three methods in a single pipeline makes it harder to fool than any single approach on its own.
PythonOpenCVscikit-learnReceiptZero — Paperless Receipt Platform (In Development)
Building a QR-based digital receipt system for Pakistani pharmacies and retail chains. The QR encodes a receipt summary directly for offline customers, and a URL for the full itemized receipt when internet is available. Delivers receipts via WhatsApp Business API, SMS, and JazzCash/Easypaisa push. Monetized through PKR subscription tiers with Safepay integration, targeting FBR-compliant merchants.
Node.jsExpressFirebaseTwilioWhatsApp Business API
Career preferences
Useful for recruiters: location, job types, salary range, and availability.