Developed a deep learning model to classify MRI scans into four categories (Glioma, Meningioma, Pituitary, No Tumor) using TensorFlow and Streamlit. Built an interactive web app for real-time classification and visualization, enhancing interpretability with saliency maps.

Built a web application for automated transcription and translation of audio content, leveraging OpenAI's Whisper model and Groq's API. Enables users to upload or record audio, segment transcripts with timestamps, and convert non-English segments to English in real-time. The application provides downloadable results, real-time status updates, and robust error handling for a seamless user experience.

Designed and implemented a web application using React and Firebase to help users manage their pantry items efficiently. The application allows users to add, edit, and delete items, as well as search for items within their pantry. Each item can be accompanied by an image for better visualization. Leveraging the Groq SDK, the app provides recipe suggestions based on the current pantry contents, encouraging users to make the most of their ingredients. The project showcases skills in front-end development, real-time data handling with Firestore, and cloud storage management with Firebase, resulting in a comprehensive tool for efficient meal planning.

Built a machine learning-based web application using Streamlit and Python to predict customer churn in the banking sector. Integrated multiple models, including XGBoost, Random Forest, and K-Nearest Neighbors, to provide churn probability for individual customers. The application offers customer-specific explanations using OpenAI's API and generates personalized emails with incentives to encourage customer retention, based on prediction insights.