Ai-driven Malicious Url Detection Using Machine Learning, Deep Learning, And Secure Full-stack Web Integration With Ci/cd Pipeline

• Author(s):

  Dr. Ravindra Krishna Chandar | Keerthi Kumar A | Satheesh Kanna S | Bavithran C | Nikesh S

• Keywords:

  Malicious URL Detection; Phishing Detection; Machine Learning; Deep Learning; Convolutional Neural Network; Recurrent Neural Network; Feature Extraction; Cybersecurity; CI/CD Pipeline.

• Abstract:

  The Pervasive Adoption Of Digital Platforms Has Precipitated A Concomitant Escalation In Phishing Attacks, Wherein Malicious Uniform Resource Locators (URLs) Serve As The Primary Vector For Credential Theft, Financial Fraud, And Malware Dissemination. Conventional Detection Paradigms, Including Blacklist-based Filtering, Signature Matching, And Rule-based Heuristics, Are Demonstrably Insufficient Against Zero-day Attacks And Polymorphic Phishing Campaigns That Continuously Mutate To Evade Static Defenses. This Paper Presents An Integrated Framework For Real-time Malicious URL Detection Leveraging A Multi-layered Artificial Intelligence Architecture Comprising Logistic Regression (LR), Decision Trees (DT), Random Forest (RF), Support Vector Machines (SVM), Convolutional Neural Networks (CNN), And Recurrent Neural Networks (RNN) With Long Short-Term Memory (LSTM) Cells. A Comprehensive Feature Engineering Pipeline Extracts URL-lexical, Domain-metadata, Content-structural, And Behavioral Attributes From Heterogeneous Data Sources Including Kaggle And PhishTank Repositories. The Proposed System Is Deployed Within A Django-based Full-stack Web Application Integrated With The VirusTotal API For Real-time Threat Intelligence Augmentation. A Secure Continuous Integration And Continuous Deployment (CI/CD) Pipeline Ensures Automated Testing, Vulnerability Assessment, And Deployment Integrity. Experimental Evaluation Demonstrates That Ensemble And Deep Learning Models Achieve Superior Classification Performance Across Accuracy, Precision, Recall, F1-score, And ROC-AUC Metrics Compared To Legacy Detection Methods. The System Addresses Identified Research Gaps Including Concept Drift Resilience, Dataset Imbalance, And Adversarial Robustness. Future Directions Encompass Transformer-based Architectures, Federated Learning, And Explainable AI (XAI) Integration.

---

Other Details

• Paper id:

  IJSARTV12I5105414

• Published in:

  Volume: 12 Issue: 5 May 2026

• Publication Date:

  2026-05-20

Download Article