I teach undergraduate and graduate courses on computer systems. Slides, assignments, and schedules are available on course website.
This course follows one AI system from the agent loop down to the GPU. Students start by using agent frameworks, then build their own, measure performance, and optimize on top of commercial APIs. They then deploy open-source models themselves and work through the underlying systems layers, with an emphasis on defining success metrics, accounting for every token and second, and measuring improvements at each layer.
A comprehensive examination of both the hardware and software components of modern computer storage systems, covering the layered architecture from individual disks and SSDs to block storage, file systems, and distributed storage systems, with an emphasis on efficiency, performance, and reliability. The first half is lecture-based; the second half features student paper presentations and guest lectures from industry experts.