Integrating Large Language Models with MLOps Observability for Attack Surface Reduction
International Journal of Global Innovations and Solutions · 2025
DOI: 10.63412/a4t8kg68 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Global Innovations and Solutions, 2025. Verfasst von Chhaya Gunawat. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Machine learning systems are increasingly being deployed in mission-critical environments, yet their attack surface is expanding due to complex CI/CD pipelines, distributed deployment, and lack of proactive observability. This paper proposes a novel integration of Large Language Models (LLMs) with MLOps observability frameworks to enhance security posture. By leveraging LLMs for real-time anomaly detection, incident reasoning, and adaptive response within MLOps pipelines, the framework aims to reduce exploitable vulnerabilities while maintaining system performance and compliance. We present an architecture where LLMs act as intelligent security co-pilots, continuously correlating logs, telemetry, and model metrics to detect adversarial activities and misconfigurations. Experimental evaluation demonstrates improved detection of adversarial injection, misconfiguration drift, and pipeline-based exploits, while significantly lowering response latency. This research highlights how LLM-augmented observability can evolve MLOps pipelines into self-defensive systems with reduced attack surface.
Zitieren
APA
Gunawat, C. (2025). Integrating Large Language Models with MLOps Observability for Attack Surface Reduction. International Journal of Global Innovations and Solutions. https://doi.org/10.63412/a4t8kg68
Vancouver
Gunawat C. Integrating Large Language Models with MLOps Observability for Attack Surface Reduction. International Journal of Global Innovations and Solutions. 2025. doi:10.63412/a4t8kg68
BibTeX
@article{Gunawat2025,
author = {Gunawat, Chhaya},
title = {Integrating Large Language Models with MLOps Observability for Attack Surface Reduction},
journal = {International Journal of Global Innovations and Solutions},
year = {2025},
doi = {10.63412/a4t8kg68}
}
EUVIMED