Supervisor-Skills

Why we included this project

Most of what makes a good advisor good never gets written down, so it is striking to see someone actually try. An assistant professor at HKUST (Guangzhou) turned a decade of publishing and reviewing at top data and AI venues into a two-part package: a handbook that walks through judging paper quality, generating ideas, writing, making figures, and handling post-submission review, and a set of skills (idea evaluator, deep research, intro drafting, paper polishing, figure design) that instruct an LLM step by step. Graduate students who rarely get timely advisor feedback can use the handbook as a reference checklist, and teams building research copilots can treat the skill definitions as worked examples of encoding editorial judgment into repeatable instructions. The project is a curated curriculum rather than a deployable service, which is what makes it useful: the methodology is readable, copyable, and easy to adapt.