AI-AUGMENTED DEVSECOPS: AUTOMATING THREAT DETECTION AND COMPLIANCE IN CLOUD-NATIVE PIPELINES USING TELEMETRY AND POLICY-AS-CODE

Abstract

The increasing complexity of cloud-native infrastructure necessitates the automation of security governance. This study presents an AI-augmented DevSecOps model that integrates telemetry data from tools like Datadog, Certificate Manager, and Linkerd with real-time policy validation using Open Policy Agent (OPA). The system enables automatic detection of configuration drifts, potential CVEs, and security policy violations across Kubernetes clusters and CI/CD pipelines. Using anomaly detection algorithms trained on infrastructure logs, the framework supports predictive incident response and auto-remediation workflows. A case study from a production-grade SOC2-compliant platform shows how the model ensures continuous compliance and reduces mean time to detect (MTTD) vulnerabilities by over 70%. The framework demonstrates significant improvements in security posture while reducing operational overhead by 65% and achieving 99.2% accuracy in threat detection with minimal false positive rates.