AI-Powered Autonomous Remote Access and Scheduling System to High-Performance Computing Resources

This research introduces an innovative AI-driven system for managing and facilitating remote access to High Performance Computing (HPC) resources. Leveraging the Django framework and cutting-edge machine learning models, the system addresses critical challenges prevalent in academic HPC environments: inefficient resource allocation, prolonged access approval delays, and a lack of user-friendly management interfaces. The proposed system’s key component include an AI-powered Lab Manager (ALM) that automates the request review process, significantly reducing approval times. Additionally, a computer-specific chatbot, developed using a fine-tuned RoBERTa-base model from Hugging Face, assists users with common queries and troubleshooting. The chatbot, trained on a custom dataset of 600 question-answer pairs, achieved an F1 score of 98.1% and an average loss of 0.135. The implementation features a user-friendly interface providing real-time access to available computing resources, instant booking capabilities, and personalized support through the AI-driven chatbot and WhatsApp integration. After implementing comprehensive security measures, the system integrates with AnyDesk for direct timely access, and it can be readily adapted to work with other remote access tools to secure and manage access effectively. For administrators, the backend system offers comprehensive resource management tools, enabling data-driven decision-making for optimized resource allocation based on specific project requirements. This innovative system significantly enhances productivity, and effective utilization of resources, and advances the mission of fostering impactful research, ultimately benefiting students, faculty, and the broader academic community.