IntelliHireX Employment Intelligence Hub: Detecting Fake Job and Internship Scams by Combining a Linear SVM Text Model with Company, Email-Domain and Document Checks

Online job portals have made hiring easier, but they have also made it easier for scammers to post fake jobs and internships that collect personal details, bank information or “training fees” from job seekers. Many fake adverts now read like professional ones, so keyword filters and text-only classifiers often miss them. A scammer can copy a genuine company’s advert word for word and change only the contact email, which a text model cannot detect. This paper presents IntelliHireX Employment Intelligence Hub, a hybrid framework that checks a job offer from several sides instead of trusting the text alone. A TF-IDF and Linear Support Vector Machine (SVM) model reads the posting. Separate modules check the company against Indian Ministry of Corporate Affairs (MCA) records, the recruiter’s email and website domain, the company’s official career page, suspect identifiers that can be reported to the National Cyber Crime Reporting Portal (NCRP), and fee-related clauses in offer letters. A transparent 100-point score combines these signals into a Low, Medium or High risk level and lists the reasons. On the public EMSCAD dataset of 17,880 postings, the text model reached 99.11% accuracy, 94.34% precision, 86.71% recall, an F1-score of 90.36% and a ROC-AUC of 0.9908 on a held-out test set of 3,576 postings, with all preprocessing fitted on training data only. The registry, domain and document modules have not yet been evaluated on labelled data. The system is deployed as a FastAPI service with a web interface.