Abstract
This manuscript represents a transformative approach to High-Performance Computing (HPC) operations and support through a Large Language Model (LLM)-based chatbot. The proposed system is designed to autonomously resolve up to 70% of commonly raised tickets, significantly reducing the operational workload and response times. By utilizing advanced fine-tuning methodologies on domain-specific HPC documentation, the chatbot ensures high accuracy, context relevance, and adaptability to complex user queries. The study emphasizes the pivotal role of prompt engineering in maximizing the chatbot's effectiveness, showcasing tailored strategies to craft precise, task-oriented prompts. We also highlight systematic evaluation methods, including accuracy metrics, usability testing, and real-world performance benchmarks, to assess and refine the chatbot's operational performance and impact. Furthermore, the framework underscores the scalability of LLM-based solutions for broader HPC challenges, such as resource allocation, job scheduling, and systems administration. This work not only enhances operational efficiency but also sets a foundation for integrating intelligent automation in HPC workflows, driving innovation and elevating the user experience in computational environments.