P&C insurers are increasingly pressured to identify and effectively predict risk. While traditional methods, such as actuarial models and manual assessments, are effective for identifying patterns in large-scale policy and claims data, they struggle to capture complex patterns, like resistance curves. This paper examines how predictive risk modelling can be implemented in practice using Guidewire DataHub and Power BI Embedded Analytics. Power BI is used for interactive visualization and real-time decision support, whereas Guidewire Data Hub is utilized as a centralized platform for storing and managing structured insurance data. It utilized structured data from claim history, underwriting attributes, policy details, and customer profiles to build a predictive model. Machine learning algorithms, such as Random Forest and Logistic Regression, were then applied to classify policyholders as High, Medium, or Low risk after preprocessing and feature selection. Standard metrics (accuracy, precision, recall, ROC-AUC) were used to evaluate model performance. The Random Forest classifier achieves an accuracy of 84% and identifies high-risk profiles most effectively. It then integrated these predictions with Power BI dashboards, allowing underwriters and analysts to explore risk at both the individual and portfolio levels. The study illustrates how building data platforms that integrate machine learning and embedded analytics facilitates more innovative underwriting, fraud detection and pricing. In a competitive, data-driven insurance environment, the ability to turn raw insurance data into actionable insights provides significant operational and strategic value.

The integration of large language models (LLMs) into mobile applications introduces new vectors for security and privacy vulnerabilities. This study proposes an automated framework for systematically testing LLM-enabled mobile apps, focusing on identifying potential threats such as prompt injection, data leakage, unauthorized access, and adversarial manipulation. The approach combines dynamic analysis, static code inspection, and machine learning-based anomaly detection to evaluate app behaviors in real-time. Our method ensures scalability and efficiency across diverse mobile platforms and LLM configurations. Results demonstrate significant improvements in detection rates and response times compared to conventional manual testing. This work aims to bridge the gap between AI innovation and secure mobile deployment, promoting trust in AI-integrated ecosystems.

Cloud platforms are now central to the management of healthcare supply chains, offering faster information exchange, scalability, and improved coordination. Yet, the very features that make them attractive also bring new privacy and security risks. Sensitive data such as patient records, procurement transactions, and shipment details move across a web of hospitals, suppliers, distributors, and service providers. This complexity raises concerns over unauthorized access, data leaks, and compliance with regulations like HIPAA and GDPR, particularly when information crosses national borders.

Common problems remain stubbornly persistent misconfigured storage systems, uneven use of encryption, and overly broad access privileges have been implicated in several breaches. Traditional safeguards, while helpful, are often too rigid for rapidly changing supply chain environments. More advanced ideas, including blockchain audit logs and federated identity systems, show potential but still face barriers in scale, integration, and oversight.

The pandemic made these tensions visible: cloud tools helped organizations track protective equipment and medicines in real time, but incidents such as the ransomware attack on Ireland’s Health Service Executive also exposed how vulnerable these systems can be.

In response, this paper puts forward a multi-layer privacy framework tailored to cloud- based healthcare supply chains. It combines encryption, adaptive access controls, blockchain-enabled transparency, and automated compliance monitoring. Findings from the study indicate that this layered approach reduces exposure to breaches, improves accountability across stakeholders, and strengthens alignment with regulatory standards. Taken together, the framework balances operational efficiency with the trust and confidentiality that healthcare delivery depends upon.

Purpose: The migration to microservices has substantially enhanced agility but simultaneously introduced critical complexities in managing inter-service communication and maintaining system-wide reliability. Traditional integration and End-to-End (E2E) testing are often too slow, brittle, and resource-intensive for continuous delivery pipelines. This article systematically examines Consumer-Driven Contract (CDC) testing, specifically through the implementation of the PACT framework, as a paradigm-shifting solution to assure inter-service reliability in distributed architectures.

Methodology: This paper develops a comprehensive theoretical framework for CDC and analyzes its technical workflow, focusing on artifact generation, mock service behavior, and provider verification. A conceptual comparative analysis is employed to contrast the PACT-based CDC approach against conventional testing strategies, utilizing key performance indicators such as pipeline execution time, integration defect leakage, and deployment frequency as metrics for success. We conduct a deep dive into the systemic role of the Pact Broker in governing contract evolution and enabling organizational autonomy.

Findings: CDC testing, particularly when formalized with PACT, is associated with a significant shift-left of integration defect discovery, leading to demonstrably faster and more stable CI/CD pipelines. The mechanism facilitates the independent evolution of services by establishing a machine-readable, shared understanding of the API interface. Crucially, the approach mitigates the coordination overhead and infrastructural cost associated with E2E environments.

Originality: This work contributes an academic synthesis of PACT's integration into the full development lifecycle, highlighting its capacity to enable high-velocity, high-assurance distributed systems, and identifying key future research avenues, including the integration of AI-driven tools.

Purpose: This article examines the architectural paradigm shift initiated by the service mesh in cloud-native microservices environments, focusing on its efficacy in enhancing system performance, security, and operational capability. The study provides a comparative analysis of prominent mesh implementations, specifically addressing literature gaps concerning resource overhead mitigation and operational complexity in advanced topologies.

Methodology: A conceptual and technical comparative analysis framework is employed, detailing the separation of concerns between the Data Plane (sidecar proxy) and the Control Plane (policy and configuration). The methodology conceptually evaluates the trade-offs introduced by leading meshes (e.g., Istio and Linkerd) across key functional areas: traffic management, mTLS-based Zero Trust security, and distributed observability.

Findings: The service mesh is demonstrated to be a foundational enabler of sophisticated resilience and security, particularly in implementing Zero Trust principles, which is associated with a reported reduction in successful lateral movement attacks. However, this infrastructure layer introduces significant performance overhead and operational complexity. Future architectural evolution is strongly associated with mitigation strategies, including kernel-level optimization (eBPF) and the movement toward multi-mesh federation to support large-scale, heterogeneous, and geographically distributed deployments. The increase in control plane development focusing on multi-cluster features confirms this trajectory.

Originality: This work synthesizes current technological trends and academic gaps, identifying the critical need for formal security verification and AI/ML-driven solutions to manage the cognitive load of mesh telemetry.

With the rapid proliferation of DevOps practices across software-intensive organizations, Continuous Integration and Continuous Deployment (CI/CD) pipelines have become the linchpin of software delivery processes. However, as deployment frequency and pipeline complexity grow — particularly when incorporating AI-enabled applications — reliability challenges and security vulnerabilities have concurrently escalated. This paper synthesizes and extends existing research to propose a unified framework that integrates AI-driven predictive modeling for failure prevention and continuous security testing within CI/CD workflows. Drawing on industry data from the 2023 state of DevOps, academic studies on predictive failure models (Patel, 2019; Enemosah, 2025), research on AI-enabled DevOps challenges (Crnkovic et al., 2020; Joseph et al., 2024), and established DevSecOps best practices (Shajadi, 2019; Brás, 2021; Jammeh, 2020; Rangnau et al., 2020; Deegan, 2020), the framework offers a detailed architecture and methodological roadmap for organizations aiming to achieve both speed and safety. Through a structured literature synthesis, thematic analysis, and conceptual modeling, findings reveal that combining predictive analytics with automated security checks can significantly enhance pipeline stability, mitigate failure-related downtime, and reduce vulnerability exposure. The discussion explores theoretical implications, practical constraints, and future research directions. This integration is posited as essential for next‑generation DevOps environments, especially in enterprises deploying AI-intensive applications.

Context: As Artificial Intelligence (AI) systems increasingly automate high-stakes decisions in healthcare, employment, and law, the opacity of "black box" models presents significant ethical and practical challenges. While deep learning offers superior predictive performance, its lack of transparency can obscure algorithmic bias and degrade user trust.

Objective: This article critiques the current landscape of Explainable AI (XAI) and fairness mechanisms, proposing an integrative framework that aligns algorithmic complexity with human cognitive limitations and ethical standards.

Methodology: We conducted a comprehensive synthesis of literature regarding XAI visualization, cognitive load theory, algorithmic bias detection, and open-source fairness toolkits. The analysis focuses on the intersection of technical interpretability (e.g., Concept Bottleneck Models, TCAV) and human-computer interaction (HCI).

Results: The review identifies a critical gap between mathematical explainability and user comprehension. High-fidelity explanations often increase cognitive load, paradoxically reducing user confidence. Furthermore, while technical bias mitigation tools exist, they are often ill-equipped to handle contextual nuances in domains like nephrology and recruitment.

Conclusion: We argue that realistic individual recourse and concept-based explanations are superior to simple feature attribution for fostering trust. Future AI development must prioritize "Open Social Innovation" and iterative usability testing to ensure systems are not only accurate but also intelligible and legally robust under frameworks like the EU’s data protection initiatives.

Background: Rapid advances in artificial intelligence (AI) have accelerated transformation across sectors—especially healthcare, retail supply chains, and financial services—producing both demonstrable operational gains and novel systemic risks (Khemasuwan & Colt, 2021; Reynolds, 2024). The convergence of generative models, federated learning, and large-scale analytics requires new frameworks that integrate technical architectures, governance, and domain-specific constraints.

Objective: This article synthesizes multidisciplinary evidence from recent applied studies, industry reports, and methodological reviews to present a coherent conceptual and operational framework for deploying AI in high-stakes domains—healthcare and finance—while accounting for data governance, model robustness, and organizational adoption processes. The aim is to move beyond descriptive summaries to produce actionable methodological guidance suitable for researchers, practitioners, and policy makers.

Methods: We conduct a structured, theory-driven synthesis of the supplied literature, mapping empirical findings and conceptual contributions onto three layers: (1) data and infrastructure (including database management and federated learning), (2) algorithmic systems (including large language models and domain-adapted generative AI), and (3) governance and organizational pathways (including compliance, risk management, and human–AI collaboration). Each layer is examined for technical requirements, failure modes, mitigation strategies, and measurable outcomes, with claims tied to the provided sources.

Results: The synthesis highlights common enablers—scalable data architectures, domain-specific model fine-tuning, and federated approaches for privacy-preserving learning—and recurring challenges—data quality, distribution shift, regulatory complexity, and the skills gap among professionals (Batra, 2018; Kaur et al., 2024; Shounik, 2025). Domain-specific vignettes illustrate how AI improves diagnostics and operational throughput in healthcare while simultaneously demanding robust validation and clinical governance (Cleveland Clinic, 2024; Mayo Clinic Press Editors, 2024; Daley, 2024). In finance, generative AI and advanced analytics enable faster due diligence and anomaly detection but require layered controls to prevent model-induced fraud and systemic concentration risks (Sergiienko, 2024; Paleti et al., 2021).

Conclusions: Realizing AI’s promise necessitates integrated architectures that combine scalable databases, privacy-preserving training (federated learning), domain-aligned model evaluation, and explicit human oversight. Implementation roadmaps must prioritize data hygiene, incremental validation, workforce reskilling, and governance protocols that reconcile business agility with safety and compliance. We offer a comprehensive operational checklist and research agenda to guide next-phase deployments and empirical evaluation.