Year: 2025
Type: article
Abstract: Chaos engineering has emerged as a critical practice to enhance the resilience and fault-tolerance of distributed systems, but its application in Large Language Models (LLMs) is still in its infancy. This paper outlines the framework for chaos engineering in LLM-based architectures, addressing key challenges and strategies for enhancing their robustness. The study presents methodologies for introducing controlled failure experiments into the LLM pipeline to ensure continued reliability and performance under unexpected conditions, making LLM systems more resilient to real-world scenarios. Keywords include chaos engineering, large language models, robustness, and fault injection.
Source: Vivekananda journal of research.
Author: Ankush Sharma
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4412468464