Effectively Using LLMs for Low-Level Systems

Catalog description

Exploratory, lab-based course (1-3 units) on using large language models to write, simplify, and rigorously check low-level hardware and OS code, via closed-loop experiments, custom instrumentation, and meta-circular tools that test and refine themselves. Labs cover topics such as closed-loop overclocking, logic- and timing-analysis for digital protocols, concurrency checking with instruction-level control, and synthesizing datasheets and drivers for undocumented hardware using LLM-driven experiments. Enrollment is invitation-only.