I'm a BS Artificial Intelligence student focused on building practical AI systems and turning machine learning models into usable software products.
My main areas of interest are:
- 🧠 Machine Learning
- 👁️ Computer Vision
- ⚡ Applied Artificial Intelligence
- 🐍 Python
- 🔌 AI Model & API Integration
- 🧩 Production-Oriented AI Systems
- 🚀 Model Deployment & Delivery
I enjoy taking AI beyond notebooks and experiments — building end-to-end AI systems, APIs, integrations, and practical AI products.
🎯 Current Focus: AI/ML Engineering + Computer Vision + production-oriented Applied AI.
An AI-assisted university timetable and clash-resolution system designed to help students, faculty, coordinators, and administrators manage scheduling conflicts safely and efficiently.
The AI architecture uses a CatBoost learning-to-rank model to prioritize already-eligible timetable-resolution candidates.
Deterministic safety rules remain authoritative for room, faculty, section, enrollment-backed student conflicts, and other hard scheduling constraints.
The ML model cannot override hard constraints, and the application falls back to deterministic ranking if the model is unavailable.
- 🧠 CatBoost
research-v1ranking model - 🛡️ Deterministic safety-gated AI workflow
⚠️ Clash detection and resolution candidate generation- 🎓 Student, faculty, coordinator, and admin workflows
- 🔐 JWT authentication and role-based access control
- 🔒 Argon2 password hashing
- 📅 CSV/XLSX timetable import
- ↩️ Controlled timetable changes with audit history and rollback
- 🗄️ SQLAlchemy + Alembic
- 🐘 PostgreSQL support
- 🧪 Pytest, Vitest, Playwright, and PostgreSQL smoke testing
- ⚙️ GitHub Actions CI
- 🐳 Docker / Docker Compose support
- 📱 Capacitor Android client
- 📚 AI evaluation, API contracts, documentation, and release artifacts
Python FastAPI CatBoost SQLAlchemy Alembic PostgreSQL React TypeScript Capacitor Docker GitHub Actions
✅ Status: Completed and released on GitHub. Designed for reproducible local evaluation and real timetable import. A continuously hosted cloud backend is not required.
An AI-powered application designed to recognize Pakistani banknotes and provide accessible feedback to users.
- 🧠 Pakistani banknote recognition using RT-DETR
- 📷 Camera and image-upload recognition
- 🔊 English & Urdu voice feedback
- 🤲 Hands-free scanning
- 📳 Vibration feedback
- 📱 Android integration using Capacitor
- ⚡ FastAPI inference backend
- ☁️ Remote AI inference through ModelScope
Python PyTorch Ultralytics OpenCV RT-DETR FastAPI React TypeScript Capacitor Android
An AI-powered resume analysis and optimization platform designed to help users evaluate and improve resumes.
- 📊 ATS-style resume analysis
- 🎯 Resume ↔ Job matching
- ✍️ Resume improvement recommendations
- 📑 PDF & DOCX generation
- 👁️ OCR and resume parsing
- 🧠 NLP and semantic similarity
- 💾 Resume management
- ☁️ Streamlit deployment
Python Streamlit spaCy scikit-learn Sentence Transformers OpenCV OCR SQLite
Currently developing Pur-Aman Pakistan, an AI-powered safety application as my university Final Year Project.
The project is being developed as a production-oriented AI system involving:
- 🤖 AI / ML capabilities
- ⚡ Backend APIs
- 🗄️ Database architecture
- 🔐 Structured backend development
- 📱 Application integration
- 🚀 Deployment-oriented architecture
Python FastAPI SQLAlchemy SQLite Alembic AI/ML
🚧 Status: Currently under active development.
PyTorch • Ultralytics • scikit-learn • OpenCV • spaCy • Sentence Transformers • CatBoost
OpenCV • RT-DETR • YOLO • PyTorch
FastAPI • SQLAlchemy • REST APIs • SQLite • PostgreSQL • Alembic
React • TypeScript • JavaScript • Capacitor • Android
Git • GitHub • Pytest • GitHub Actions • Docker • VS Code • Android Studio
Streamlit Cloud • ModelScope • GitHub Releases
✅ Completed → UniTime-AI — AI-Assisted Timetable & Clash Resolution
🛡️ FYP → Pur-Aman Pakistan
👁️ Focusing → AI/ML Engineering + Computer Vision + Applied AI
⚡ Developing → Production AI Systems, FastAPI & Model Integration
🚀 Goal → Build AI Products That Solve Real Operational Problems