AI Code Assistance for HPC on Palmetto
AI Code Assistants for HPC on Palmetto is a hands-on workshop focused on using AI code assistants effectively in real HPC workflows (Slurm jobs, logs, environments, and performance constraints). We’ll briefly compare tool options (ChatGPT web interface, GitHub Copilot, Claude Code, and others), but the workshop is centered on OpenAI’s Codex CLI as a practical “pair-debugger” and productivity tool for Palmetto-specific work. Through guided exercises, you’ll practice using an assistant to iterate on failed job submissions, interpret error output, and make targeted improvements to correctness and efficiency -- without treating the assistant as an authority.
Live Workshop
Session #1 for Summer 2026
Prerequisites
Participants should have completed the Palmetto self-guided onboarding training and be comfortable working in a terminal (e.g., SSH’ing to Palmetto, navigating directories, editing files, and running basic commands). Workshop participants must also have active Clemson ChatGPT Edu accounts, and have installed OpenAI Codex (either the CLI or the desktop app or both) on the laptop you'll bring to the workshop. The workshop exercises will use Python scripts; familiarity with Python will help with completing the exercises but is not required.