Register to join our August 26, 2026, webinar, “AskALCF: An Intelligent HPC User Support System in the Age of AI,” to learn how AI can streamline HPC user support by bringing together facility documentation, scheduler manuals, programming-model references, and accumulated support knowledge into a unified, continuously improving system.
HPC user support is challenging because the information users need is often distributed across many sources, and solutions developed through support tickets do not always make their way back into the documentation. AskALCF is designed to close that gap by exposing a shared knowledge base spanning ALCF, OLCF, NERSC, and LLNL documentation, along with PBS and Slurm scheduler resources and CUDA, oneAPI, SYCL, and HIP-ROCm programming references, through a chatbot interface accessible to both users and AI coding assistants.
Beyond answering questions, AskALCF powers an autonomous support pipeline that can triage difficult queries and incoming tickets, reproduce issues on real systems, compare results against the knowledge base to identify documentation gaps, and generate human-reviewed pull requests to improve support materials. In this way, the system becomes more effective with every resolved case.
In this webinar, Huihuo Zheng of the Argonne Leadership Computing Facility will present the AskALCF architecture, demonstrate its capabilities live on real ALCF systems, and share lessons learned about building trustworthy AI-driven support for leadership-class computing environments.