SecureCheck | Devpost

Scam or Not a Scam?
Trick or Treat?
Explanation Screen
End Screen
Precision and Recall on our Model
Transaction Frequency

Inspiration

Our team was inspired by the increasing sophistication of fraud tactics and the need for more effective prevention methods. We wanted to create a solution that not only detects fraud but also empowers users to recognize and avoid scams themselves.

What it does

SecureCheck offers a three-step approach to fraud prevention through three key components: an AI-driven fraud detection model for credit card transactions, an interactive quiz to educate users on identifying fraudulent communications, and advanced audio and text analysis for detecting scam calls and messages. We use these tools together to create a world where scams and fraud will become a thing of the past.

How we built it

We utilized our own custom model based on DNNs trained on public, MIT-licensed available data of credit card transactions. We combined a variety of state-of-the-art data analysis techniques in order to fix various imbalances and data imperfections in order to train a model with a 96% validation accuracy! We also utilized various industry technologies to create our beautiful web-stack, including Flask, HTML, CSS, JS, Ajax, and Python.

Challenges we ran into

We faced challenges in fine-tuning our AI model to achieve high recall without sacrificing precision. Additionally, creating engaging educational content for the quiz required iterative testing and feedback from users to ensure clarity and effectiveness. Also, syntax errors. So many syntax errors. JavaScript is a fun language, innit?

Accomplishments that we're proud of

Extremely high model accuracy. UI built for both spooky season and non-spooky season (Corporate ver.).

What we learned

Various data analytics practices Hackathons are tiring :((( Frontend development

What's next for SecureCheck

We plan to expand our features by incorporating more educational resources, improving our AI model with continuous learning, and exploring partnerships with financial institutions to broaden our reach. We could utilize LLMs to generate fraudulent messages to tune the learning experience for users who may have a tendency to fall victim to certain kinds of scams. We aim to make SecureCheck the go-to tool for anyone looking to enhance their fraud prevention knowledge and capabilities, ensuring financial safety through education and cutting-edge detection technology.