Phinding Nemo

Phinding Nemo
Extension tools
Extension Login/Sign Up

Inspiration

We were inspired by the rapid evolution of technology, the need to protect students' information online, and the power of machine learning to quickly and efficiently address complex challenges. Our goal was to harness these advancements to create a solution that keeps students safe from online threats.

What it does

Phinding Nemo is a browser extension that scans the content of emails to assess potential threats. It utilizes a neural network trained to distinguish between safe and scam emails, identifying dangers before they can harm the user. The extension then provides a clear evaluation of the email, guiding users to make informed decisions and ensuring a seamless and secure online experience.

How we built it

Phinding Nemo was developed in three main parts:

Browser Extension: We created a Google Chrome extension using JavaScript, HTML, CSS, and MongoDB. This extension allows users to activate the tool, log in, and scan emails for potential threats, providing a seamless user experience.
Website: We built a website using React, HTML, CSS, and JavaScript to showcase our project. The site provides an overview of Phinding Nemo, details about the tools we used, and instructions for downloading the browser extension.
Neural Networks: To analyze email content, we implemented neural networks using Python and frameworks like PyTorch. We trained our model on open-source datasets, enabling it to accurately distinguish between scam and safe emails. We prioritized minimalistic design across all sections to make the interface user-friendly and intuitive while maintaining effective functionality.

Challenges we ran into & What we learned

Throughout this project, we faced several challenges, particularly in learning how to create and implement a neural network capable of processing data and adapting to new examples. We conducted extensive research and experimentation with various models, frameworks, datasets, classification techniques, and evaluation methods. This process pushed us to learn new concepts, including Python libraries and frameworks that facilitate our workflow.

Additionally, we explored front-end tools to effectively connect the different sections of our project using Flask and enhance the user interface with React. We also delved into web scraping, as our original idea involved analyzing the HTML content of emails. This experience taught us valuable lessons about integrating diverse technologies and adapting our approach based on the project's needs.

Accomplishments that we're proud of

We are proud to have created a tool that enhances online security by providing users with feedback on potential threats in emails. We recognize that this is a widespread issue affecting many people. Additionally, we take pride in the rapid learning and implementation that allowed us to integrate various features—such as the browser extension, web app, and neural network—into a seamless, cohesive security solution.

What's next for Phinding Nemo

The next steps for Phinding Nemo involve expanding our training dataset to strengthen the neural network and experimenting with different features, algorithms, and probabilities to enhance efficiency. Additionally, we aim to expand our database to provide users with more comprehensive information during their sessions, further improving the overall user experience.

Built With

Submitted to

NewHacks 2024
- Winner Best Domain Name from GoDaddy Registry

Created by

I worked as full stack developer, I worked on the creation of the Chrome extension, that reads and outputs whether an email is phishing, or scam.

Rafael Jabbour
I worked on the RNN algorithm, learning Pytorch to develop a Neural Network that embeds individual characters, with an embedding, GRU, and Linear layer. I trained the network on over 10,000 email strings using Google Colab, and achieved an accuracy of over 90%.

NPinzin Pinzin
Grace Hao
Matéo Tomeho