Hi, I'm Yashmitha Desai

Architected a real-time Digital Twin platform featuring interactive 3D visualization (Three.js) and live telemetry via WebSockets...

Architected a real-time Digital Twin platform featuring interactive 3D visualization (Three.js) and live telemetry via WebSockets, enabling immersive remote monitoring of HVAC dynamics and building systems. Deployed advanced ML pipelines (LSTM & Autoencoders) trained on the ASHRAE dataset to forecast 24-hour energy loads and detect system anomalies, driving data-led decisions for energy efficiency and retrofitting. Integrated a GenAI assistant powered by Llama-3.2-3B to interpret complex building metrics through natural language, allowing facility managers to instantly query status and receive optimization suggestions.

An intelligent matchmaking system that pairs students with complementary strengths and weaknesses for collaborative learning...

An intelligent matchmaking system that pairs students with complementary strengths and weaknesses for collaborative learning, using Natural Language Processing to analyze student profiles and optimize study groups. Features a Hybrid Matching Engine combining NLP Semantic Analysis with a Knowledge Graph to understand both context and skill relationships. Provides dual matching scores—"AI Match" (sentiment/context) and "Graph Match" (structural skill overlap)—for every pairing. Built with a modern dark-themed UI featuring glassmorphism effects and dynamic score badges. Includes a REST API powered by FastAPI with automatic validation and Knowledge Graph integration, plus a Project Relevance Hub to find projects that fit skills using semantic and structural relevance.

A comprehensive relational database system for managing BMTC's operational data including drivers, buses, routes, shifts, and complaints...

A comprehensive relational database system for managing BMTC's operational data including drivers, conductors, buses, routes, shift schedules, complaints, and accident reports. Built with a normalized database schema using PostgreSQL to eliminate redundancy and ensure ACID compliance for data integrity. The system features a full-stack architecture with Node.js/Express backend using Sequelize ORM for seamless database operations, and a React/Vite frontend styled with Tailwind CSS for responsive, user-friendly interfaces. Includes Data Flow Diagrams (DFDs) representing system processes, supports essential administrative functions like driver management, route planning, shift scheduling, complaint tracking, and accident analysis. The centralized platform improves operational efficiency, enhances data consistency, and provides better decision-making support for transport authorities.

An interactive React app that demonstrates sorting algorithms using real-time stock market data...

An interactive React app that demonstrates sorting algorithms (Bubble, Merge, Quick Sort) using real-time stock market data. It features adjustable visualization speeds, live API data integration with Yahoo Finance, and educational tools including algorithm pseudocode, performance metrics, and a "Guess the Algorithm" challenge.

Built an advanced gardening app using Android Developers and Arduino for hardware integration...

Built an advanced gardening app using Android Developers and Arduino for hardware integration. Integrated sensors to monitor key parameters like soil moisture and temperature. Automated gardening suggestions based on sensor data and weather predictions. Combined software and hardware solutions for efficient, data-driven automation.

Built a Flask-based web app to detect phishing using Levenshtein distance, Jaccard similarity, and entropy scoring...

Built a Flask-based web app to detect phishing using Levenshtein distance, Jaccard similarity, and entropy scoring. Visualized suspicious domain clusters using Neo4j to identify brand-targeting patterns. Designed a Chrome extension for real-time domain checking with a visual risk meter and redirect suggestions. Integrated bulk domain checker, SSL certificate status, and WHOIS lookup for enriched threat analysis.

Built an AI-driven system to shortlist football candidates based on performance metrics...

Built an AI-driven system to shortlist football candidates based on performance metrics. Utilized NumPy and Pandas for data analysis and preprocessing. Visualized candidate performance using Matplotlib and Seaborn for insights. Trained machine learning models with Scikit-learn for candidate evaluation. Developed a dynamic frontend using React.js for user interaction. Integrated a Node.js backend for efficient data handling and model deployment. Secured user authentication by storing login details in MongoDB. Successfully hosted the website on Streamlit.

Designed a responsive and interactive website using HTML, CSS, and JavaScript with ML-powered fraud detection...

Designed a responsive and interactive website using HTML, CSS, and JavaScript. Developed a secure login system using PHP and hosted locally with XAMPP. Trained a machine learning model in Python to detect credit card fraud using real-world datasets. Integrated the ML model with the website backend using Flask for seamless prediction functionality. Ensured smooth client-server communication with RESTful API endpoints. Focused on user security and data integrity in all transactions.

Designed an interactive yoga website using HTML, CSS, and JavaScript with secure login functionality...

Designed an interactive yoga website using HTML, CSS, and JavaScript. Implemented secure login functionality using PHP. Deployed the website locally using XAMPP for testing and demonstration. Focused on user-friendly design and functionality.