The proliferation of online recruitment platforms has correspondingly amplified the incidence of fraudulent job postings, posing grave risks to employment seekers in terms of financial exploitation, identity theft, and psychological harm. Conventional approaches to identifying deceptive job advertisements have relied predominantly on heuristic keyword matching and rule-based filtering, which exhibit substantial deficiencies with respect to adaptability, scalability, and explainability. This paper presents an Explainable Artificial Intelligence (XAI)-powered Smart Job Scam Detection System that integrates Hybrid Natural Language Processing (NLP) techniques with behavioural feature engineering to classify job postings as genuine or fraudulent with high accuracy. The proposed system employs Term Frequency–Inverse Document Frequency (TF-IDF) vectorization, suspicious keyword detection, recruiter email domain verification, company profile consistency analysis, and ensemble machine learning classification using XGBoost and Logistic Regression. The Explainability module leverages SHapley Additive exPlanations (SHAP) to provide transparent, human-interpretable reasoning behind every prediction. Additional functionalities include Optical Character Recognition (OCR)-based screenshot analysis, batch CSV prediction, scam probability scoring, risk level classification, and an interactive visualization dashboard. Experimental evaluation on the Kaggle Fake Job Postings dataset demonstrates an overall accuracy of 97.4%, with a Precision of 96.8%, Recall of 95.9%, and F1-Score of 96.3%, outperforming baseline methods. The system presents a robust, transparent, and scalable solution to combat online recruitment fraud.

Keywords — Explainable AI, Fake Job Detection, NLP, TF-IDF, XGBoost, SHAP, OCR, Behavioural Feature Engineering, Fraud Detection, Cybersecurity.