Project Experience
Academic Data Analytics Platform
| Python, Dash Plotly, MySQL, MongoDB, Neo4j, Aiven, Render
| website
|

- Full-Stack Development: Developed a comprehensive web-based analytics dashboard enabling prospective graduate applicants to explore academic programs, compare universities, and identify prominent researchers through interactive data visualizations
- Cloud Infrastructure: Deployed multi-database architecture utilizing Aiven for MySQL, MongoDB Atlas, and Neo4j Aura for scalable backend services; implemented seamless hosting on Render for real-time updates
Mobile Weather Application | Android Studio, Java
- Android Development: Engineered a feature-rich weather application with secure user authentication, customizable UI themes, real-time weather data integration via Google Maps API, and AI-powered Q&A functionality using Gemini API
- Quality Assurance: Implemented comprehensive test automation with Espresso framework, developing instrumented tests to validate core functionalities and ensure application stability across diverse user scenarios
C++ Systems & Game Development | Visual Studio, C++
- Console Game Programming: Developed 2D console-based games in C++ including Tic-tac-toe with AI opponent and Gomoku with configurable board dimensions and winning conditions, implementing object-oriented design patterns and algorithms
- File System Implementation: Built a modular file system supporting core operations (create, delete, open, close) with password-based security; developed custom shell commands (ls, rm, cat, copy) for efficient terminal-based file management
Computer Vision & Neural Networks | Jupyter, Python, TensorFlow/Keras
- Image Processing & Feature Engineering: Preprocessed 45,000+ paperclip images with noise reduction techniques; engineered features by extracting RGB pixel intensities to optimize model training performance
- Neural Network Design: Designed and trained a 4-layer neural network using TensorFlow/Keras framework to predict paperclip quantities from extracted features, achieving RMSE ≤ 2.0 and 92% validation accuracy