The "Aegis" Framework: A Multi-Cloud, Fault-Tolerant MLOps Architecture for Real-Time Financial Decisioning and Regulatory Compliance
DOI:
https://doi.org/10.15662/IJEETR.2025.0706031Keywords:
MLOps, Finance, Regulation, Multi-Cloud, Architecture, ComplianceAbstract
The paper provides the findings of a quantitative study on Aegis Framework, a multi-cloud MLOps framework that is aimed at financial institutions. The results point out that the framework is much more beneficial on the availability, speed of the fail-over, and precision of governance in the scenario of working with real-time decision workloads. The cross-cloud routing did not break the inference services whenever cloud failures were detected, and Mean Time to Recovery was less than two seconds. During all stress tests, checking of governance was not lost and metadata logs were not lost. Peak loads high throughput and low latency was obtained as well. These findings confirm that Aegis offers high reliability and resilience as well as compliance performance to the modern financial systems.
References
[1] Eken, B., Pallewatta, S., Tran, N. K., Tosun, A., & Babar, M. A. (2025). A Multivocal Review of MLOps Practices, Challenges and Open Issues. A Multivocal Review of MLOps Practices, Challenges and Open Issues. https://arxiv.org/pdf/2406.09737v2
[2] Liu, C., Tan, R., Wu, Y., Feng, Y., Jin, Z., Zhang, F., Liu, Y., & Liu, Q. (2024). Dissecting zero trust: research landscape and its implementation in IoT. Cybersecurity, 7(1). https://doi.org/10.1186/s42400-024-00212-0
[3] Watson, H. J., & Larson, D. (2024). MLOps. International Journal of Business Intelligence Research, 15(1), 1–22. https://doi.org/10.4018/ijbir.358916
[4] Pourmajidi, W., Zhang, L., Steinbacher, J., & Erwin, T. (n.d.). A reference architecture for governance of cloud native applications. In Toronto Metropolitan University, Toronto, Canada. https://arxiv.org/html/2302.11617v2
[5] Liu, C., Tan, R., Wu, Y., Feng, Y., Jin, Z., Zhang, F., Liu, Y., & Liu, Q. (2024b). Dissecting zero trust: research landscape and its implementation in IoT. Cybersecurity, 7(1). https://doi.org/10.1186/s42400-024-00212-0
[6] Madugula, S. R. P. (2024). MLOPS, MODEL RISK MANAGEMENT (MRM) & GOVERNANCE FOR ACTUARIAL ML. In TIJER - INTERNATIONAL RESEARCH JOURNAL, TIJER - INTERNATIONAL RESEARCH JOURNAL (Vol. 11, Issue 4) [Journal-article]. https://tijer.org/tijer/papers/TIJER2404262.pdf
[7] Joshi, S. (2025). Model Risk Management in the era of Generative AI: challenges, opportunities, and future directions. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.5206477
[8] Dreibholz, T., & Mazumdar, S. (2022). Towards a lightweight task scheduling framework for cloud and edge platform. Internet of Things, 21, 100651. https://doi.org/10.1016/j.iot.2022.100651
[9] Moskalenko, V., & Kharchenko, V. (2024). Resilience-aware MLOps for AI-based medical diagnostic system. Frontiers in Public Health, 12, 1342937. https://doi.org/10.3389/fpubh.2024.1342937
[10] Azad, M. A., Abdullah, S., Arshad, J., Lallie, H., & Ahmed, Y. H. (2024). Verify and trust: A multidimensional survey of zero-trust security in the age of IoT. Internet of Things, 27, 101227. https://doi.org/10.1016/j.iot.2024.101227
