ScamShield : AI-Powered Job Scam Email Detector

A machine learning system that detects and prevents job scam emails to protect users from fraud, identity theft, and emotional distress—boosting trust in digital hiring.

Comment

🔍 Inspiration The inspiration for this project came from a personal experience—one of our team members was nearly a victim of a sophisticated job scam. What seemed like a legitimate offer from a reputable organization turned out to be a fraud, exposing the emotional and financial risks that countless job seekers face every day. This incident fueled our motivation to build a robust, intelligent system that helps protect others from falling into similar traps.

🛠️ What We Built We developed JobSentinel, a machine learning-based system that classifies job offer emails as legitimate or fraudulent. We trained and evaluated multiple models including Logistic Regression, SVM, and Random Forest on a hybrid dataset built from real, synthetic, and crowdsourced job emails. The final model, a tuned Random Forest classifier, achieved 96.7% accuracy and excellent recall, making it highly effective at flagging scam emails.

📊 Key Features

  • Synthesized scam indicators (e.g., non-HTTPS links, bait phrases, ID info requests)
  • SHAP-based explainability for model transparency
  • Hybrid features from email content, metadata, and sender behavior
  • Real-time detection-ready architecture

💡 What We Learned We learned that the quality of features often outweighs model complexity. We also gained hands-on experience with class imbalance challenges, hyperparameter tuning, feature engineering, SHAP interpretability, and agile team collaboration.

🚧 Challenges

  • Balancing a highly imbalanced dataset (only ~6% scam emails)
  • Creating realistic synthetic scam data without introducing noise
  • Ensuring our model could generalize well to new, unseen scam tactics
  • Making the system explainable and suitable for real-time integration

🌱 Impact JobSentinel goes beyond spam filtering. It supports social sustainability by protecting vulnerable groups, economic sustainability by reducing identity fraud, and contributes to UN SDG 8 by promoting ethical recruitment practices.

Built With

Share this project:

Updates