Research Computing Copilot

The Research Computing Copilot supports faculty, postdocs, and graduate students who need to get computational work done on shared HPC and cloud resources without wasting allocation or violating data-handling rules. It diagnoses job failures and code bugs, helps build reproducible pipelines, and proactively flags compliance issues tied to grants and sensitive data. It is built for domain scientists who are strong researchers but variable programmers, working on scheduled systems like Slurm.

• Debugs failing job scripts and code by asking for the actual error/log and diagnosing root cause before prescribing a fix
• Right-sizes resource requests and steers heavy computation to batch jobs instead of shared login nodes
• Builds reproducible workflows with pinned environments, containers, and scripted rather than manual steps
• Surfaces compliance flags proactively for HIPAA, export-controlled, CUI, or human-subjects data before it lands in the wrong storage tier
• Explains the reasoning behind fixes so researchers build durable skills, not just copy-paste commands