Lectures

Slides and the papers discussed in each class are listed below.

• 

  Course introduction

  tl;dr: Course overview and research themes in advanced computer networks.

  Slides & links: PDF
• 

  Lecture 1 Internet with Agents

  tl;dr: Coding agent workloads, WebSocket-based communication, and Model Context Protocol (MCP).

  Slides & links: PDF

  Reading list

  1. TraceLab: Characterizing Coding Agent Workloads for LLM Serving
  2. Speeding up agentic workflows with WebSockets in the Responses API , OpenAI
  3. Model Context Protocol
• 

  Lecture 2: LLM serving

  tl;dr: KV cache memory management, inference scheduling, and disaggregated architectures for efficient LLM serving.

  Reading list

  1. [SOSP'23] Efficient Memory Management for Large Language Model Serving with PagedAttention (slides)
  2. [OSDI'24] Taming Throughput-Latency Tradeoff in LLM Inference with Sarathi-Serve (slides)
  3. [OSDI'24] DistServe: Disaggregating Prefill and Decoding for Goodput-optimized Large Language Model Serving (slides)
  4. [FAST'25] Mooncake: Trading More Storage for Less Computation — A KVCache-centric Architecture for Serving LLM Chatbot (slides)
  5. [NSDI'26] Agentix: An Efficient Serving Engine for LLM Agents as General Programs (slides)
  6. [NSDI'26] JITServe: SLO-aware LLM Serving with Imprecise Request Information (slides)
  7. [NSDI'26] ServeGen: Workload Characterization and Generation of Large Language Model Serving in Production (slides)
• 

  Lecture 3: Networking for AI

  tl;dr: In-network aggregation, collective communication algorithm synthesis, and performance simulation for distributed machine learning.

  Slides & links: Introduction to Collective Communication

  Reading list

  1. [NSDI'21] SwitchML: Scaling Distributed Machine Learning with In-Network Aggregation (slides)
  2. [NSDI'23] TACCL: Guiding Collective Algorithm Synthesis using Communication Sketches (slides)
  3. [NSDI'25] SimAI: Unifying Architecture Design and Performance Tuning for Large-Scale Large Language Model Training with Scalability and Precision