PhishGuard AI

🚀 Inspiration

Every second, phishing emails slip through filters and land in the inboxes of unaware users. But what if we could build a bodyguard — one that reacts instantly, silently, intelligently?

That’s how PhishGuard AI was born:
A serverless defender that wakes up the moment a suspicious email lands, scans it with AI precision, and shields the user — before damage is done.

🔧 What It Does

• 📨 Intercepts inbound emails using Amazon SES
  
• ⚡ Triggers AWS Lambda on email receipt
  
• 🧠 Analyzes content using phishing heuristics and AI (Amazon Bedrock)
  
• 📦 Logs verdicts to DynamoDB, stores emails in S3
  
• 🚨 Sends alerts if phishing is detected
  

All of this, without any servers or manual intervention.

🔨 How We Built It

• Verified a test email with Amazon SES
• Built a Python AWS Lambda to process incoming emails
• Integrated with Amazon Bedrock for intelligent phishing detection (Claude or Titan)
• Used S3 for archiving, DynamoDB for metadata, and SNS for real-time alerts
• Deployed & tested in Ireland (eu-west-1) SES-supported region

🧱 Challenges We Faced

• SES only supports inbound triggers in a few regions — had to migrate setup
  
• Parsing SES email payloads accurately
  
• Tight Lambda IAM permissions
  
• Crafting fast, reliable phishing heuristics
  

But the reward? A truly reactive, real-world security solution — zero infrastructure.

🧠 What We Learned

• Deep end-to-end serverless pipeline using AWS
  
• Email parsing, SES triggers, and AI-integration in real time
  
• Security automation with no Ops burden
  

Built With

• devpost
• github
• python-3.12-aws-lambda-amazon-ses-(inbound)-amazon-s3-amazon-dynamodb-amazon-sns-amazon-bedrock-(llm-analysis)-cloudwatch-markdown