Muhammad.

Muhammad Umer

Full-Stack Developer & AI Engineer Specializing in Scalable Real-Time Applications

Overview

A Computer Science student at LUMS with extensive experience building AI-driven platforms and real-time multiplayer systems using React, Flutter, and FastAPI. He excels at integrating advanced LLMs like Gemini and Claude into secure, production-ready applications that solve complex behavioral and legal challenges.

Lahore, Pakistan

Project Portfolio

ACADEMIC

Present

MindBloom: AI-Driven Study Behavior Modeling

Lead Architect / Developer

Architected a client–server system spanning a React/Vite web app, Flutter mobile companion, and Express REST API, unifying focus tracking, gamification, flashcards, and AI-driven study recommendations on Firestore and Supabase Storage. Built a Python FastAPI microservice to serve trained regression and classification models predicting study session productivity and completion probability from behavioural signals, using the StudentLife, EdNet, and ASSIST datasets alongside survey data from 50+ undergraduate students. Drove personalized study recommendations by scoring candidate focus durations (15–60 minutes) against combined productivity/completion predictions; layered in the Google Gemini API for AI-generated flashcards and conversational study support. Conducted STRIDE threat modeling across 20 use cases, surfacing gaps that led to OWASP-aligned controls (Firebase RBAC, AES-256 at rest, TLS in transit, rate limiting, WAF) hardening the authentication, storage, and API layers. Built and validated an automated test suite (Jest, Playwright, Firebase Emulator) covering 10+ critical workflows, confirming state sync consistently under 5 seconds and AI recommendation latency under 15 seconds ahead of release.

ReactViteFlutterExpressREST APIFirestoreSupabase StoragePythonFastAPIGoogle Gemini APIFirebase RBACAES-256TLSWAFJestPlaywrightFirebase Emulator

PERSONAL

Present

Qatra Hayat: Real-Time Blood Donation Platform

Full Stack Developer

Built a real-time blood donation app in Flutter and Firebase, matching donors to recipients instantly based on blood type, location, and request urgency. Modeled the Firestore database schema and backend logic across users, donors, recipients, blood drives, and messaging entities to support scalable, concurrent application workflows. Shipped a location-aware alert pipeline (Firebase Cloud Messaging + Google Maps API) that delivered emergency notifications and donor search results in under 10 seconds – critical in time-sensitive requests where delays directly affect patient outcomes. Added the Google Gemini API for AI-powered donation guidance, alongside a donor credibility rating system and real-time in-app messaging. Secured the app end-to-end with Firebase Authentication and Firestore Security Rules for role-based access control, validated through automated integration testing on the Firebase Emulator Suite.

FlutterFirebaseFirestoreFirebase Cloud MessagingGoogle Maps APIGoogle Gemini APIFirebase AuthenticationFirebase Emulator Suite

PERSONAL

Present

Know Your Nikahnama: Legal Awareness Platform

Developer

Built a bilingual legal awareness platform in Next.js and React, making the Pakistani Nikahnama’s clauses understandable through an interactive conversational interface. Designed a rule-based NLP engine mapping free-text English and Urdu queries to relevant Nikahnama clauses via keyword matching, pattern recognition, and automatic Urdu script/numeral detection. Layered the Anthropic Claude API on top as a clause-scoped fallback reasoning engine, ensuring graceful fallback to the rule-based engine during API outages. Implemented conversation management, response pagination, and a fully digitized in-app Nikahnama reference, streamlining legal information retrieval for end users.

Next.jsReactNLPAnthropic Claude API

PERSONAL

Present

ColorGrid Multiplayer Game

Backend & Game Logic Developer

Built a real-time multiplayer web game on the MERN stack and Socket.IO, delivering synchronized, low-latency gameplay across concurrent users. Designed an event-driven client–server architecture handling matchmaking, room allocation, turn management, live score updates, and game state synchronization. Wrote backend game logic, socket event handling, and anti-cheat validation, keeping multiplayer sessions consistent and abuse-resistant across concurrent rooms. Optimized WebSocket communication and server-side event processing to reduce perceived input latency and improve gameplay responsiveness.

MERN stackMongoDBExpress.jsReactNode.jsSocket.IOWebSockets