Research — Neural Network Pruning
IEEE CONECCT 2025 — a chaos-stimulated method that identifies and prunes causal weights to shrink models while preserving accuracy.
Hover to prune the network
Read on IEEE XploreIEEE CONECCT 2025 — a chaos-stimulated method that identifies and prunes causal weights to shrink models while preserving accuracy.
Hover to prune the network
Read on IEEE XploreThis very site — animated, responsive and component-driven, with a hand-crafted design system and real-time GitHub integration.
Move your cursor — draw a trail
View on GitHubAndroid music streaming app with Firebase auth, real-time playlists and a custom player UI built from scratch.
Hover to scrub the track
View on GitHubMachine-learning analysis of behavioral signals — data preprocessing, feature engineering and classification in Python.
Drag the decision line
View on GitHubA collection of projects built with passion, exploring animation, design, machine learning, and mobile development.
Year: 2025 | Venue: IEEE CONECCT | Pages: 1–6 | DOI: 10.1109/CONECCT65861.2025.11306624
A peer-reviewed research paper on efficient neural network pruning, co-authored and presented at the 2025 IEEE International Conference on Electronics, Computing and Communication Technologies (CONECCT). The method identifies the causal weights that drive misclassifications and prunes them via chaos-stimulated weight updating — reducing model size for resource-limited devices while preserving accuracy and explainability.
Year: 2023 | Tech Stack: HTML, CSS, JavaScript, React
An interactive portfolio web application showcasing skills, projects, and experience. Built with React for component-based architecture and state management. Features smooth animations, responsive design, and integrations with GitHub to dynamically display project information. This very site you're exploring now is the evolved version of this original portfolio.
Year: 2022 | Tech Stack: Java, XML, Android Studio, Firebase
A feature-rich music streaming application for Android built from scratch. The app implements core music player functionality with a modern UI, user authentication via Firebase, and seamless playlist management. Designed to demonstrate proficiency in Android development, XML layouts, and real-time database integration.
Year: 2024 | Tech Stack: Python, Pandas, NumPy, Jupyter Notebook
A machine learning project analyzing behavioral and textual patterns to identify potential depression indicators. Built with Python data science libraries, it processes structured datasets, performs statistical analysis, and applies classification models. Demonstrates expertise in data preprocessing, feature engineering, and ML model evaluation using Jupyter for exploratory analysis.