ChefGPT - An Inclusive and Interactive Community App with an Enhanced Recommendation System

ChefGPT is an AI-powered web application that provides personalized recipe recommendations while prioritizing accessibility for visually impaired users. It helps users discover meals tailored to their preferences, dietary restrictions, and sustainability goals while also addressing challenges faced by differently-abled individuals when using existing culinary platforms. ChefGPT uses a hybrid recommendation system that combines content-based filtering (CBF), collaborative filtering (CF) with K-means clustering and popularity-based ranking, and neural collaborative filtering (NCF). This approach ensures a diverse and personalized selection of recipes. The platform features a dynamic social front-end integrated with a FastAPI-based backend and Firebase for real-time data synchronization. It was tested using a synthetic dataset of 5000 Indian recipes and simulated user interactions. It achieved strong results: an RMSE of 0.1146, MAE of 0.1229, and an 87% success rate in handling cold-start users. The system also maintains a 22% serendipity rate and a Jaccard similarity diversity score of 0.41, ensuring that recommendations balance relevance and novelty. Accessibility is at the core of ChefGPT, with voice commands, screen reader compatibility, and automatic image descriptions enhancing usability. Looking ahead, future updates will integrate real-time ingredient data, introduce Transformer-based recipe generation, and link dietary data for a more holistic, health-conscious experience.