I am COO at Dev Hatch Labs, a Machine Learning and Full Stack developer passionate about building intelligent, real-world applications.
I specialize in combining AI (ML, NLP, Computer Vision) with modern web technologies to create scalable and practical systems.
π₯ Currently building AI-powered chatbots, LLM applications, RAG systems, and automation tools for real-world business solutions.
π Top 10 out of 23+ Teams!
AI-powered virtual assistant for University of Layyah with bilingual support (English/Urdu) for students, applicants, and visitors.
Tech Stack: Flask Groq AI Llama 3 70B MySQL BeautifulSoup Vercel HTML/CSS/JS
Key Features:
- π Role-Based Responses (Students, Applicants, Visitors)
- π Bilingual Support (English + Urdu)
- π€ Voice Input (Web Speech API)
- π Text-to-Speech
- π΄ Offline Mode with pre-loaded answers
- π Rate Limiting (30 req/min)
π Live Demo | π Repository
π Healthcare ML Challenge @ FAST NUCES Lahore
Predict whether a healthcare member will incur high medical costs (> $30,000) in the next calendar year using historical administrative data.
Tech Stack: Python LightGBM XGBoost Scikit-learn Pandas NumPy
Key Features:
- Ensemble: LightGBM + XGBoost (Soft Voting)
- Recall: 0.825 | F1: 0.616 at threshold 0.35
- 10+ Engineered Features from multi-table data
- Handled 11% positive class imbalance
Tech Stack: Python LangChain RAG Streamlit ChromaDB
- AI assistant for University of Layyah using RAG architecture
- Natural language query processing
- Integration with university data
- π GitHub | π Live Demo
Tech Stack: Python XGBoost LightGBM Scikit-learn Pandas
- Predictive model for healthcare cost optimization
- Achieved 0.825 recall in identifying high-risk patients
- Ensemble approach with threshold tuning
- π GitHub
Tech Stack: Flask MySQL Bootstrap jQuery Chart.js
- Complete food ordering and management system
- Real-time order tracking and analytics dashboard
- Role-based access for admins and customers
- π GitHub
Tech Stack: TensorFlow Keras OpenCV Tkinter
- Real-time American Sign Language recognition
- Trained on custom dataset with 88% accuracy
- User-friendly GUI for accessibility
- π GitHub
Tech Stack: Python YOLOv8 OpenCV Tkinter
- Real-time product detection and inventory tracking
- Achieved 92% accuracy on custom retail dataset
- Automated stock monitoring with visual alerts
- π GitHub
| Project | Tech Stack | Status |
|---|---|---|
| π High-Cost Patient Prediction | Python, XGBoost, LightGBM, Scikit-learn | β Complete |
| π ASL Sign Language Recognition | TensorFlow, Keras, OpenCV, Tkinter | β Complete |
| π Smart Retail Shelf Monitoring | Python, YOLOv8, OpenCV, Tkinter | β Complete |
| π¬ Sentiment Analysis (RNN) | TensorFlow, Keras, NLP | β Complete |
| Project | Tech Stack | Live Demo |
|---|---|---|
| β Smart Cafeteria | Flask, MySQL, Bootstrap, jQuery, Chart.js | GitHub |
| π€ Digital Debate Judge | Flask, SQLite, Bootstrap | β |
| π TechNest | React, Vite, JavaScript | β |
| Project | Tech Stack | Live Demo |
|---|---|---|
| π SwiftEats | HTML, CSS, JavaScript | Demo |
| π LuxEstate | HTML, Tailwind CSS, JavaScript, Chart.js | Demo |
| Project | Tech Stack | Live Demo |
|---|---|---|
| π Apex Dashboard | HTML, CSS, JS, FullCalendar, Chart.js | Demo |
| π° AURUM Finance | HTML, CSS, JS, Chart.js | Demo |
| π¦ Nimbus Weather | HTML, CSS, JS, OpenWeather API | β |
β Building intelligent systems with AI + Full Stack development β
