SmartCart
SmartCart is a mobile recipe recommendation and grocery-planning app developed as my undergraduate capstone project. It combines grocery-list scanning, personalized recipe discovery, and cart-building into a single workflow. Users can upload or photograph a grocery list to automatically populate their cart, receive recipe recommendations based on dietary restrictions and ingredient preferences, and add the ingredients required for a selected recipe directly into their cart before checking out.
Why I Created It
SmartCart was developed as a two-semester undergraduate capstone project exploring how grocery shopping and meal planning could be made more connected and efficient. Rather than treating recipe discovery, grocery-list management, and cart building as separate tasks, our team designed a single mobile experience that could move users from planning what they wanted to eat to preparing the groceries needed to make it. I was especially interested in incorporating an AI-driven recommendation component that reflected my concentration in artificial intelligence.
How I Built It
The app was built with a React Native frontend and a Node.js, Express.js, and MongoDB backend. OCR was used to extract grocery items from uploaded or photographed lists and automatically populate the user's cart. I developed the recipe recommendation system using OpenAI embeddings and an SVM-based ranking approach to match users with recipes based on dietary restrictions, ingredient preferences, and other profile information. Recipe data was retrieved through the Spoonacular API, and selected recipes could automatically populate the cart with their required ingredients. The application also included account registration, two-factor authentication, and supporting REST API workflows. Development followed a Scrum process across four monthly sprints, and I served as Scrum Master for two sprints, coordinating documentation, progress tracking, and team productivity.
Challenges
The primary challenge was balancing the project's feature scope against a four-month development window. Our original product vision included more functionality than the team could realistically implement within the semester, which required us to prioritize the core shopping and planning workflows. The recipe recommendation engine was particularly important to me because I wanted the final project to include a substantive AI component. Building and integrating that system late in the development timeline required narrowing the scope carefully while still delivering the feature before the capstone deadline.
Impact
SmartCart resulted in a working capstone prototype that integrated OCR-based grocery-list scanning, personalized recipe recommendations, automated ingredient-to-cart workflows, authentication, and a complete mobile shopping experience. The project also gave me hands-on experience developing an AI recommendation system within a larger full-stack application while working in a structured Scrum team environment.