ai systems seminar

course details

info increasingly, natural language question answering systems and database-style analytics systems are converging: with overlapping missions across agentic "deep research" systems and multimodal data lake integration systems.

this seminar will explore the common ideas, principles, and architectures behind high-performance data-centric AI systems. we will study how to develop, efficiently train, and deploy advanced systems like large-scale generative models and compound AI systems across heterogeneous hardware and distributed platforms.

topics include modern distributed data systems, parallel and distributed ML/AI frameworks for training and inference, modular AI systems, memory + I/O-aware algorithm design, retrieval-augmented generation, planning frameworks, and agentic orchestration.
instructors
zack ives zives [at] cis
varun jana jvarun [at] cis
ryan marcus rcmarcus [at] cis
meeting time: wed + fri @ 1:45-3:14p
place: agh 315