This project addresses the need for visually branded digital assets by implementing a premium QR Code Generator that transcends standard monochromatic matrix codes. Developed a responsive front-end application using Vanilla JavaScript and optimized canvas rendering engine pipelines. Implemented Dynamic Theme Presets and granular customization controls for dot geometries, corner styles, and adaptive color palettes, featuring Runtime Logo Merging with cross-origin handling. Enabled high-fidelity asset exportation in PNG and SVG formats to ensure print and digital scalability.

This is a Python-based human-computer interaction (HCI) application that enables touchless control of your operating system using real-time hand gestures. Leveraging OpenCV and MediaPipe for computer vision, the system tracks hand landmarks with a standard webcam and maps specific finger movements to desktop commands. Users can glide the mouse cursor, trigger clicks with a pinch gesture, control system volume with a fist, and switch browser tabs with a rapid swipe. Designed with dynamic smoothing and safety fail-safes, this project offers a seamless, futuristic, and accessible interface for navigating your workspace without a mouse or keyboard.

Developed an interactive DNA sequence alignment application using React and TypeScript to demonstrate the Needleman-Wunsch algorithm for global sequence alignment. The platform facilitates dynamic recurrence matrix calculation, supports configurable parameters for match/mismatch/gap scoring, and parses .fasta files automatically for comparison. Created a high-fidelity visual interface displaying interactive Traceback paths and alignment contrast layouts to provide researchers and students with intuitive insights into biological homology calculation.

Built an end-to-end, AI-powered solution for predictive environmental analytics, tailored for real-time air quality monitoring. Designed a robust data pipeline that cleans and interpolates noisy time-series sensor data to train an XGBoost Regression model leveraging recursive sliding-window forecasting. Developed a performant web dashboard featuring live modular panels and Chart.js visuals streaming continuous 24-hour look-back triggers. Deployed the system using FastAPI to yield immediate decision-support capabilities for climate and smart-city metrics.

Built a Deep-Fake Verification System utilizing a PyTorch-based ResNet18 model to classify digital manipulation in media streams. Integrated Grad-CAM backpropagation to generate forensic activation heatmaps, providing visual transparency into model decision-making by highlighting anomalous gradients. Custom-built a high-performance FastAPI backend to run inference dynamically and calculate real-time Trust Scores. Designed a premium Next.js Dashboard featuring drag-and-drop analysis overlays and an On-Chain Source Verifier using Solidity Smart Contracts to audit media authenticity and anchor origin metadata immutably on the ledger.

Developed a real-time Sign Language Translator that converts hand gestures into text using Computer Vision. Built using MediaPipe for high-precision landmark tracking and Scikit-learn's Random Forest classifier for fast, lightweight inference on standard CPU hardware. Implemented a Streamlit dashboard that visualizes dataset analytics, manages model training workflows, and launches live camera overlays. Created a fully automated pipeline with synthetic dataset expansion for recording, training, and deploying custom sign letters instantly.