The increasing heterogeneity of cloud, enterprise, and Internet of Things (IoT) environments raises concerns about the long-term reliability of machine-learning-based intrusion detection systems (IDSs). This study evaluates temporal robustness and cross-domain generalization using four publicly available datasets collected between 2017 and 2024. Five representative models—Random Forest, Gradient Boosting, Multi-Layer Perceptron, Autoencoder, and a lightweight 1D-CNN—are assessed under in-dataset, forward temporal, enterprise-to-IoT transfer, and dataset-agnostic evaluation protocols without retraining. In the dataset evaluation, models achieve Macro-F1 scores between 0.84 and 0.96. However, forward temporal testing reveals consistent degradation, with performance reductions reaching ΔF1 ≈ 0.20–0.27 when models trained on 2017 enterprise traffic are applied to IoT datasets from 2023 to 2024. Under cross-domain transfer, Macro-F1 decreases to 0.69–0.78, and benign false-positive rates increase up to 0.30, indicating substantial sensitivity to traffic distribution shifts. Tree-based ensemble models show comparatively lower degradation (≈6–23%) and reduced performance variance across datasets. Semantic feature analysis further indicates that flow intensity and temporal activity features exhibit higher cross-dataset stability than protocol-dependent indicators. These findings demonstrate that IDS robustness in evolving Internet environments depends strongly on evaluation methodology and feature stability, highlighting the need for generalization-oriented assessment strategies.
Security operations face increasing complexity due to evolving cyber threats and regulatory compliance requirements. Traditional incident response methods often suffer from delayed triage and limited readiness for compliance audits. This paper presents an AI-driven Digital Twin framework designed to simulate and analyze security threats, enabling faster incident triage and improved compliance readiness. The proposed system leverages serverless data engineering pipelines, real-time streaming analytics, and AI/ML-driven anomaly detection models integrated within a sandboxed Digital Twin environment. By orchestrating telemetry ingestion (e.g., logs, IoT sensor data, and network events) with scalable cloud native services, the framework supports high-frequency threat simulations without impacting production systems. Experimental evaluations using synthetic and public cybersecurity datasets (CICIDS2017, UNSW-NB15) demonstrate up to 40% faster compliance reporting, 35% faster incident triage, and improved anomaly detection accuracy compared to both SIEM and SOAR workflows. Unlike traditional cyber ranges or static compliance tools, this work integrates AI-driven anomaly detection with Digital Twin modeling and serverless pipelines, enabling elastic, real-time what-if threat simulations and automated compliance posture reporting validated against SIEM and SOAR baselines.
The fast pace of transformation of systems heavy on transactions, such as the digital, has created the need to intelligently, yet at scale, represent, process and act upon huge amounts of information in real time. The proposed framework discussed in this paper switches the traditional streaming architecture into a predictive, automated cloud AI workflow. The proposed model simultaneously provides a decision-making improvement in high-throughput applications like finance, retail and smart infrastructure by using edge-to-cloud orchestration, real-time data pipelines and event processing through AI. The architecture connects the stream ingestion, the feature extraction, the federated learning and the predictive analytics into a unified pipeline with a minimum amount of manual interaction. There is a more significant latency management, more precise prediction, and resiliency in performing different loads of transactions. This piece highlights how cloud-native AI automation has the potential to change reactive streaming backends into proactive intelligent platforms, which can intelligently react and scale in dynamically changing digital environments.
We present a multi-cloud, serverless deployment framework and a complementary AI pipeline for outbreak management in large healthcare networks. The framework extends an operational rules-based system deployed during COVID-19 (15,000 employees across 295 facilities) by adding portable Infrastructure-as-Code (IaC), cost-aware scaling, and healthcare-grade security. The AI pipeline integrates facility-level forecasting (LSTM), individual risk scoring (XGBoost), anomaly detection (Isolation Forest), and contact-network analytics, with explainable AI (XAI: SHAP, LIME, counterfactuals) embedded directly into clinician workflows. Evaluation on anonymized operational logs and calibrated synthetic multi-disease streams shows improvements over (i) a rules-only production baseline and (ii) a single-vendor “siloed AI” deployment: higher predictive accuracy, lower end-to-end latency, $\sim 30 \%$ lower cost per million events, and increased clinician agreement. We explicitly delineate (A) benefits from multi-cloud portability (e.g., disaster recovery, data residency, cost control) and (B) benefits from AI augmentation. Together, these yield a portable, explainable, and cost-conscious foundation for outbreak management beyond COVID-19.
Manual security policy validation of Infrastructure-as-Code (IaC) creates bottlenecks in enterprise CI/CD pipelines, with 90% of cloud breaches involving misconfigured IaC. Traditional static analyzers struggle with evolving cloud services and custom policies. We propose a production-ready framework augmenting conventional scans with Large Language Models (LLMs) for Kubernetes, IAM, and Terraform validation. Our evaluation on 500 synthetic IaC cases shows ensemble methods achieve F1 = 0.95 at 3.1s latency. LLMs detect complex violations missed by rule-based tools, reducing manual review by 60% and maintenance by 70%. Real-world testing in Jenkins and Bamboo confirms cross-platform compatibility. We provide: (i) privacy-preserving CI/CD architecture, (ii) safeguards against prompt injection and hallucination, and (iii) phased rollout strategy for regulated enterprises balancing security and velocity.