← Back to Projects
BERT + IF Hybrid Fraud Engine

Hybrid NLP + Anomaly Detection

Guardian Recruit

Guardian Recruit: Hybrid Fraud Detection & XAI for Digital Recruitment

Fraud Detection XAI NLP Machine Learning Python Data Engineering

Relevant for: Machine Learning · Data Engineering

Master’s Capstone Research Lead | UNT DTSC 5082

February 2026 – May 2026

Project Summary

Guardian Recruit is my master’s capstone: a decision-ready fraud detection system for digital recruitment platforms. It combines natural language processing, anomaly detection, and explainability to help platforms identify suspicious job listings with clearer, more actionable risk signals.

🔑 Key Contributions

1. Architectural Leadership

Orchestrating the project roadmap and designing a dual-stream Fusion Layer to integrate linguistic patterns with statistical metadata for real-time fraud detection.

2. Hybrid Modeling

Engineering a system that cross-references BERT-based NLP features with Isolation Forest anomaly detection to identify inflated salary-to-experience ratios and AI-generated fraudulent listings.

3. Explainable AI (XAI)

Implementing SHAP-driven risk scores to provide interpretable reasoning for detections, ensuring job seekers can verify the validity of listings via economic benchmarking.

4. Robust Data Engineering

Managing the ingestion of the EMSCAD dataset and leading a web-scraping initiative to augment the model with 2026 labor market data to ensure contemporary relevance.