Creative Showcase
A curated selection of my most impactful work, ranging from complex full-stack architectures to refined frontend experiences.
CRM Travel (SNE)
An advanced, fully-fledged B2B Travel CRM and booking management ecosystem designed for travel agencies and corporate businesses. It streamlines the entire lifecycle of hotel reservations, banquet bookings, flight ticketing, and transportation logistics, featuring AI-driven quoting, dynamic task assignments, and automated vendor communications.
Key Features
- End-to-end booking workflow management for hotels, banquets, flights, and transport
- Dynamic task board with interactive drag-and-drop UI for query status tracking
- Automated email notifications and rate request generation for vendors
- AI-powered quotation builder to generate highly customizable premium quotes
- Role-based access control with comprehensive user and company administration
- Real-time analytics dashboard with key performance statistics and KPI tracking
NearbuyDukan
A hyperlocal digital ecosystem connecting consumers with local shops and healthcare institutions. Enables virtual queue management, digital billing, real-time chat, location-based discovery, and secure medical record storage—eliminating waiting time and paper-based workflows.
Key Features
- Live token-based queue tracking for customers and patients
- Location-based discovery of nearby shops and healthcare centers
- Secure medical history and document storage
- Real-time chat between users and vendors
- Digital invoice generation with automated record keeping
- QR-based instant access for token booking and payments
Weather Application
A modern React-based weather application delivering real-time weather data, 3-day forecasts, and hourly insights with geolocation support. Features a premium glassmorphism UI, responsive design, and optimized rendering for smooth performance.
Key Features
- Real-time weather data including temperature, humidity, wind speed, UV index, and pressure
- Automatic geolocation-based weather detection
- City-based global weather search functionality
- 3-day forecast with interactive hourly weather visualization
- Responsive glassmorphism UI with dynamic twilight theme
- Performance optimization using GPU acceleration and efficient rendering techniques
ChatSphere
A full-stack real-time chat application built with the MERN stack and Socket.IO. Enables one-on-one and group conversations with live messaging, user presence tracking, and secure authentication.
Key Features
- Real-time one-on-one messaging using Socket.IO
- Group chat with multiple participants
- Online/offline user presence tracking
- Secure authentication with OTP-based password reset
- Profile customization with image upload (Cloudinary)
- Lazy loading of messages for performance optimization
- Search functionality for conversations and groups
- Responsive UI with toast notifications
Accident Fatality Prediction
A machine learning-based predictive system that analyzes large-scale traffic accident data to forecast accident severity. Achieved ~85% accuracy using Random Forest and Logistic Regression, enabling data-driven road safety insights and risk assessment.
Key Features
- Built predictive models (Random Forest, Logistic Regression) achieving ~84–85% accuracy
- Performed extensive data preprocessing including cleaning, feature engineering, and encoding
- Implemented multiple missing value strategies (Mean vs Mode) for performance comparison
- Transformed temporal data into meaningful features like Time of Day
- Analyzed feature importance to identify key factors influencing accident severity
- Visualized trends and model performance using data analysis techniques
Digit Recognition System
A deep learning-powered web application that recognizes handwritten digits in real-time using a CNN model trained on the MNIST dataset. Integrated with a Django backend to deliver fast and secure predictions through an interactive web interface.
Key Features
- Real-time handwritten digit recognition using a CNN model
- Achieved high accuracy on MNIST dataset (trained on 60,000+ samples)
- Implemented image preprocessing including resizing, grayscale conversion, and normalization
- Built a responsive UI for image upload and instant predictions
- Secure backend integration using Django with CSRF protection
- Optimized model inference for fast and efficient predictions