International Journal of Intelligent Systems and Applications in Engineering

Secure AI-Driven Identity Infrastructure for Regulated Sectors

Pramod Gannavarapu — 2025-11-12

The regulated industry is facing increased threats to identity as attacks on fleeting clouds, APIs, and mobile devices increase the attack surface. To provide real-time monitoring, adaptive authentication, and auditable compliance, this paper suggests a secure and AI-driven identity infrastructure, a fusion of ML, NLP, and LLM-driven analytics with zero-trust design. The work was inspired by the fact that identity-related breaches are prevalent in most breaches, including the hack of Global data breach victims in 2023 by a total of 43% of all government breaches, which are common, particularly in healthcare, finance, and government ecosystems, where HIPAA, GDPR, and SOX control identity theft. The architecture employs authentication, authorization, entitlement, and evidence in restricted, microservice settings; expels operational telemetry into online feature stores; and implements policy-as-code in a hybrid cloud. Approaches combine scaling risk scoring with gradient-boosting session anomaly sequence models, toxic entitlement graph analytics, and federated learning to minimize data movement. Covering 100,000 sessions in four weeks, the stack outperforms rules-based baselines on metrics including accuracy, recall, F1, ROC-AUC, median arena decision latency, coverage of compliance, newly seen evidence, and false challenges. Findings show that 92 out of 100 fraud attempts were reported caught in 30 seconds (vs. a 60-second baseline), resulting in 31 fewer attempted frauds, a 19-point improvement in coverage of evidence, and the availability of decisions at 99.6% in stressful situations. Up to 35% of fraud was reduced, and the user experience was maintained at a much higher level using adaptive authentication. The study concludes by discussing governance, privacy-saving methods, and a research agenda focused on standard benchmarks, compliance, and adaptive compliance.

DOI: https://doi.org/10.17762/ijisae.v13i2s.7913

Scalable Data-Driven Engineering for High-Performance Computing & Financial Services

Santosh Durgam — 2025-11-12

The study discussions will touch upon enhancements to the efficiency of high-performance computing systems through the assimilation of scalable data engineering and semiconductor verification to improve data computing technological processes, especially in the financial services sector. The authors of this study explore how computerized gross working of applications like fraud detection, risk control, and algorithmic trading can be optimized through the integration of GPU hardware verification with AI-ready systems. The major results of the study include that the performance of the system increased by 25%, and the time spent on studying cases decreased as the proposals of the GPU made it possible, and the provisions of AI-driven hardware verification to conduct the verification were enhanced by 30%. Moreover, it also found data engineering pipelines to compute 95% success rates when handling real-time financial transactions, and with latency reduced to under <5ms. A combination of scaled, automated data streams has led to cost reductions of 15-20% per year, and the use of machine learning in semiconductor testing has shown that the decrease in tests could be up to 90%. The paper concludes that convergence of hardware validation and scalable data engineering is a key to devising AI-ready systems, capable of meeting the computational needs of the present financial services, enhancing reliability, scalability, and velocity. The integration is set to drive the real-time ability of decisions in the economic field to a greater extent.

DOI: https://doi.org/10.17762/ijisae.v13i2s.7914

Cloud-Powered Healthcare & Insurance Transformation with CRM and Advanced Analytics

Sridhar Rangu — 2025-11-12

Healthcare and insurance agencies are experiencing disjointed experiences, an increase in service demands, and stringent regulations that hinder cost, quality, and expertise. The paper presents a proposal of a cloud-native blueprint that will merge Salesforce Health Cloud, Service Cloud Voice, and Einstein Copilot with a controlled analytics stack to modernize engagement, care coordination, and claims. Data is extracted through FHIR/HL7 and APIs into a lakehouse and feature store and operated to serve risk, service, and fraud models and operationalized within a CRM component based on retrieval-augmented guardrails. They are entity resolution to a Member/Patient 360, PHI tokenization, and experiment-ready instrumentation. With expected results of 10-20% decrease in Average Handle Time, a 6-10 percentage point increase in First-Contact Resolution, a 15 percent reduction in claims cycle time, readmission AUROC >0.82, precision of SIU logged at 1k ≥0.60, and ≥99.9% logged with ≤0.1% policy exceptions, steps A/B were hired, the randomized-encouragement and threshold A/B designs are applied. Observability focuses on P95 API latency of <300 ms and ASR WER under 12% and the cost per member per month is regulated and taxed to be between $0.08–$0.25 per month. The donation will include a Remote deployable compliant reference architecture, measurement plan, connecting the model measures to business KPIs, and guardrails on fairness, safety, and reliability. The solution can be applied to federated learning, multimodal analytics, and streaming interoperability through FHIR Subscriptions and payer-to-payer APIs. Research is applicable across both payers and providers and contributes to a gradual rollout and clear governance, financial, and well-defined performance WM.

DOI: https://doi.org/10.17762/ijisae.v13i2s.7915

AI First Framework for Predictive Database Replatforming

Rahul Jain — 2025-11-12

The paper proposes the AI First Framework of Predictive Database Replatforming, which focuses on integrating the gradient boosting, graph neural networks and optimization models. The quantitative analysis was carried out on 480 workloads of enterprises to assess the results of costs, risk, and migration. The findings indicate that the AI-based system will be 45 percent more accurate in predicting ROI, cut down on downtimes by 40 percent, and will be more cost-effective by 25 percent than the conventional system. The framework is a methodical means of planning complicated database migrations, through predictive intelligence, which has quantifiable business worth, reduced operational risk, and quicker conversion to contemporary cloud-based database frameworks.

Design And Implementation of Parallel Processing Fir Filter Using Modified Booth Encoding

S. Selvakumar Raja — 2025-11-12

This project presents a Parallel Processing FIR filter or Full Parallel FIR Filter design implemented using the Modified Booth Encoder (MBE) to achieve high-speed performance. A novel hardware architecture employing fine-grained seamless pipelining is proposed, where pipeline registers are strategically placed not only between components but also across them. This ensures minimal gate delays and maximizes throughput. A precise critical path analysis at the gate level enables an optimal pipelining strategy tailored to throughput requirements. The results show that the fully-parallel FIR filter achieves very high throughput with a substantial reduction in area delay product (ADP) compared to existing systolic designs. The proposed design optimizes MBE encoding and parallel processing to achieve significant enhancements in speed, area, and power efficiency. FIR filters are crucial components in digital signal processing applications, but traditional designs often face limitations. The proposed design employs a pipelined architecture, leveraging MBE encoding to reduce computational complexity. This research investigates the potential of parallel processing and MBE encoding in FIR filter design analysing trade-offs between speed, area, and power consumption.

Task-Specific Image Enhancement for Underwater Turtle Detection and Segmentation: An Investigation on a Benchmark Dataset

Abhisheka Thumbesara Eshwara — 2025-11-20

Underwater images often present challenges related to color distortion, low contrast, and loss of texture, severely degrading the performance of computer vision models involved in ecological monitoring or autonomous systems. This work investigates the performance of various enhancement techniques based on two main downstream tasks: turtle detection using the YOLO model and turtle segmentation using SAM. A multi-species dataset of turtles was collected from public resources, and five representative enhancement schemes, including classic contrast enhancement, generative learning-based enhancement, physics-guided correction, fusion-based processing, and the proposed TOUE method, were tested. Experimental results illustrate that the effectiveness of enhancement is very task dependent. That is, TOUE had the best detection accuracy and generalization capability on both the custom dataset and the SUIM benchmark, whereas CLAHE generated the best segmentation accuracy owing to consistent local contrast refinement. No single enhancement method effectively produced the optimal outcome in these two tasks. Guided by this observation, a dual-stage pipeline has been proposed, utilizing TOUE for detection and CLAHE for segmentation, a more reliable end-to-end pipeline for underwater vision. These findings underline the necessity of choosing enhancement methods based on the downstream application and provide, for the first time, a practical framework for optimizing detection-segmentation systems in real underwater scenarios.

Resilience by Design: Site Reliability Engineering in Financial Platforms

Vasudevan Subramani — 2025-11-20

This paper examines the application of Site Reliability Engineering (SRE) principles to enhance stability, customer experience, and operational efficiency in financial platforms. Modern information systems demand extremely high availability, as even minor outages can lead to revenue loss, regulatory fines, and reputational damage. The case study demonstrates that automation and SRE practices prevented over $1 million in penalties, minimized failed transactions, and improved root cause analysis through custom ETL file management and heat maps. Additionally, the MyAccount portal was redesigned to reduce errors and improve usability, while operational improvements cleared 7,000 backlog tickets and reduced daily ticket volume to fewer than 68. Telemetry and failover automation further increased system availability to 99.95%. Findings confirm that SRE is a technical methodology rather than a customer-facing approach, enabling organizations to reduce costs, improve efficiency, and deliver services reliably. The conclusions highlight the strategic importance of SRE in fintech and its potential to shape robust, scalable, and cost-effective platforms.