This project is an AI-powered financial wellness tool that goes beyond simple expense tracking to identify psychological spending biases like "Subscription Creep" or "Payday Splurge." By combining traditional Machine Learning (Random Forest) for bias detection with the Gemini LLM for conversational coaching, it transforms raw transaction data into a meaningful dialogue about financial self-awareness.

This project implements a decentralized machine learning system designed to predict diabetic patient readmissions across 10 independent nodes without sharing sensitive raw data. Utilizing a server-client architecture containerized with Docker, the system performs federated training on daily healthcare partitions, aggregating local model weights into a robust global model while maintaining patient privacy.

This project is a full-stack web application that performs sentiment analysis on movie reviews by comparing traditional machine learning algorithms with modern deep learning architectures. It features a React-based interface where users can input text to see real-time classification results across multiple models, including Logistic Regression, SVM, and BiLSTM.

This project implements a rule-based simulation framework to optimize emergency medical dispatching by balancing immediate response times with long-term coverage. By applying targeted "exception rules" (R1 and R2) to a nearest-ETA baseline, the system protects Advanced Life Support (ALS) availability and critical zone coverage during periods of high demand or low supply.

Grading Assistant is an AI-driven platform that automates the evaluation of coding assignments by using the Gemini LLM to generate reference solutions and provide rubric-aligned feedback. It bridges the gap between efficiency and accessibility, offering instructors a human-in-the-loop dashboard to manage high-volume submissions while ensuring fair, transparent results for students.