AI Enhanced DevSecOps Pipeline for Automated Database Deployment & Security

Abstract

With rising security concerns in software systems, DevSecOps has emerged to embed security into DevOps workflows; however, this integration often challenges agility and slows delivery speed. To address this challenge, this study presents an AI-enhanced DevSecOps pipeline designed for secure and automated database deployment, leveraging machine learning for real-time intrusion detection without compromising speed. The methodology incorporates rigorous preprocessing, including feature selection, class balancing with SMOTE, and feature scaling, followed by training and evaluation of ML models such as K-Nearest Neighbor (KNN), Multi-Layer Perceptron (MLP), and Logistic Regression (LR). Experimental results demonstrate that KNN outperforms other models with an accuracy of 99.57%, precision of 99.80%, recall of 99.55%, and F1-score of 99.67. In addition, the suggested models demonstrate better performance and scalability when dealing with the complicated TII-SSRC-23 dataset when compared to current methods such as Decision Tree (DT), J48, NaïveBayes (NB), and LSTM. The best-performing model was integrated into a CI/CD pipeline. This enabled intelligent threat detection during deployment. The work shows that security can be embedded proactively into DevOps. AI helps preserve agility while strengthening cybersecurity. The framework is scalable and high-performing.