My friends kept getting lost while reading a book, so I built them an app

My friends often asked me for help when they got lost while reading a book. This prompted me to create a simple React application to assist them with their inquiries. The AI model powering the app was running locally, so it wouldn't be functional once the web app was hosted online. However, I've provided screenshots and the GitHub repository for reference.

I employed the MERN stack for this project, despite my limited experience with it. The frontend was built using React, while the server and backend were constructed with Node.js and Express.js. MongoDB served as the database to store the conversations and the title of the book being read. The book title is obtained from Open Library's API.

For the AI model, I selected gemma3:1b-it-q4_K_M and integrated it with the Express server using Ollama's API. Although the model wasn't as advanced as I had hoped, I had to use it due to limited storage on my computer. Since the model was smaller in size compared to other Gemma models, I chose it and adapted my approach accordingly.

The decision to run the AI model locally with Ollama provided me with greater control over the model. I could use it as I pleased, without being dependent on a closed API. This local setup allowed for experimentation and customization to suit my needs.

Using Ollama to run the AI model offered several advantages. It enabled me to have full control over the model, which I could use whenever I wanted, without relying on a closed API. This local setup allowed me to run and experiment with the model as I saw fit. The project was recognized in two categories: Best Use of MongoDB Atlas, as all conversations were stored in a MongoDB database, and Best Use of Gemma, with gemma3:1b-it-q4_K_M being the model that answered questions about the books.