CS 395T: Advanced Topics in Systems and GenAI (Fall 2025) is a graduate-level reading and project seminar investigating how modern generative AI—large language, vision and multimodal models—interacts with computer systems. We will study:

• Systems for GenAI—scaling, serving, and optimising LLMs, diffusion and agentic workloads
• GenAI for Systems—how foundation models can design, debug or operate complex infrastructure.

By the end of the semester you will have:

• Surveyed the state of the art
• Practiced critical paper reviewing
• Explored research problems through a semester-long research project

The course is structured around lectures by the instructor Aditya Akella, guest lectures, and paper readings/presentations by the students with open discussion. Details will be posted on the Schedule tab. Students will form a project group (two or three students) and conduct a research project on systems and generative AI.

See the Logistics tab for detailed information on course organization and policies.

What We Will Cover

A tentative reading list lives on the Reading List tab and draws heavily from OSDI/SOSP, ASPLOS, SIGCOMM, NSDI, MLSys and Nature papers.

Part 1 – LLMs as the Backbone of Modern AI

• Parallel & elastic training (3D, MoE, fault-tolerance)
• Networking for multi-datacentre scale
• Low-latency, energy-aware inference & serving

Part 2 – GenAI Beyond Text

• Diffusion and video generation systems
• Retrieval-augmented pipelines and multi-agent orchestration
• Cloud/edge schedulers for multimodal workloads

Part 3 – GenAI for Systems

• AI-assisted formal verification, configuration and workload synthesis
• LLM-powered debugging, root-cause analysis & telemetry

• Time: Tuesday & Thursday 12:30 PM - 02:00 PM
• Location: GDC 6.202
• Discussion: Ed (TBD)