PhishGuard

• 

  All emails are shown by default marked as safe or unsafe. Red is unsafe and green is safe

• 

  Phishing emails are only shown after selection

• 

  Safe emails are shown here after selection of safe email option

Inspiration

Took an InfoSec course so had an added interest in cyber security and how phishing is the most common technique to launch large scale Cyber Kill Chain

What it does

Filters your emails and detects potential phishing scams

How we built it

Trained a BERT model on email data. Created a Flask API for hosting it and creating endpoints which the user frontend can call and therefore utilize it to filter emails. IMAP is used for authentication

Challenges we ran into

-Using IMAP for authenticating the user. The biggest challenge as everything else was ready and working.

Accomplishments that we're proud of

• Training BERT and creating a successful API on Flask to use it (never done the deploy on flask thing before hehe)
• The model had an accuracy of 99.4% after fine tuning

What we learned

• About IMAP protocols and how to host our model through Flask -friendship is magic :)
• how to use app passwords and OAUth

What's next for PhishGuard

• Further streamline the UI and make it more intuitive -Fix the authentication issues on IMAP
• Deploy it as a browser plugin to integrate with one's gmail

Built With

• ai
• bert
• css
• flask
• html
• imap
• python
• pytorch
• transformers