Schedule and Reading List

Date	Readings	Presenter
	Part 1: The Foundation of Research (Metaskills)
Sun 01/25/2026	Introduction + Systems Research	Marco
Wed 01/28/2026	Reading/Reviewing Papers	[video1]
Sun 02/01/2026	Tech Trends in the Datacenter	Marco [recording1]
Wed 02/04/2026	The Era of Agentic AI and What It Means for Systems Research	Marco & Zafar
Sun 02/08/2026	Research Patterns	[video2]
Wed 02/11/2026	How to Innovate + Clear/Creative Thinking	[video3]
Sun 02/15/2026	Giving a Talk and Time Management + People Skills	[video4] [video5]
	Part 2: Technical Core – Multi-Agentic Systems
Wed 02/18/2026	Foundations & Agent Architectures Goal: Understanding the “Agentic Loop” where models transition from static text generation to iterative reasoning and environmental interaction through tool-use. ReAct: Synergizing Reasoning and Acting in Language Models (ICLR ‘23) Reflexion: Language Agents with Iterative Self-Reflection (NeurIPS ‘23) Toolformer: Language Models Can Teach Themselves to Use Tools (NeurIPS ‘23)
Sun 03/01/2026	Multi-agent frameworks & Failure Modes Goal: Analyzing how different interaction patterns (conversation vs. role-playing) enable complex task completion, while identifying the unique bottlenecks and bugs inherent in collaborative LLM systems. AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation (COLM ‘24) Reflective-Agent Collaboration based on Language Language Models (NeurIPS ‘24)
Wed 03/04/2026	MetaGPT: Meta Programming for a multi-agent collaborative framework (ICLR ‘24) Why do Multi-Agent Systems Fail?
Sun 03/08/2026	Automated Design & Self-Evolution Goal: Exploring the move toward “Zero-Touch” agent engineering, where the system itself discovers optimal multi-agent topologies and training regimes without human intervention. ADAS: Automated Design of Agentic Systems (ICLR ‘25’) MAS-Zero: Designing Multi-Agent Systems with Zero Supervision
Wed 03/11/2026	MAS-GPT: Training LLMs to Build LLM-based Multi-Agent Systems TBA
Sun 03/15/2026	Programming Models & Compilers Goal: Investigating the shift from brittle manual prompting to structured programming abstractions and compilers that optimize agent logic for reliability and code execution. Executable Code Actions Elicit Better LLM Agents (ICML ‘24) DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelines (NeurIPS ‘24) DMAS-Forge: A Framework for Transparent Deployment of AI Applications as Distributed Systems
Wed 03/18/2026	No class - holiday
Sun 03/22/2026	No class - holiday
Wed 03/25/2026	No class - holiday
Sun 03/29/2026	Decentralized Coordination & Networking Goal: Rethinking internet architecture and communication protocols to support decentralized, semantic-aware agent interactions. AgentNet: Decentralized Evolutionary Coordination for MAS Multi-Agent Coordination Via Multi-Level Communication (NeurIPS ‘24)
Wed 04/01/2026	Toward an AI-Native Internet: Rethinking the Web Architecture for Semantic Retrieval Beyond DNS: The NANDA Index and Verified AgentFacts
Sun 04/05/2026	Visualization and Plotting	Marco
Wed 04/08/2026	Mid-Semester Presentations
Sun 04/12/2026	Agentic Infrastructure & Data Systems Goal: Examining how data storage and retrieval must be fundamentally redesigned to support the high-concurrency and state-management needs of autonomous agent HedraRAG: Co-Optimizing Generation and Retrieval for Heterogeneous RAG Workflows (SOSP 2025) PATCHWORK: A Unified Framework for RAG Serving (2025)
Wed 04/15/2026	Supporting Our AI Overlords: Redesigning Data Systems to be Agent-First (2025) Scaling Large Language Model-Based Multi-Agent Collaboration (ICLR ‘25)
Sun 04/19/2026	Scheduling & Serving Agents Goal: Tackling the systems-level challenges of serving LLM applications, focusing on optimizing semantic variables and fair scheduling to minimize latency in multi-agent workflow Parrot: Efficient Serving of LLM-based Applications with Semantic Variable (OSDI ‘24) Justitia: Fair and Efficient Scheduling for LLM Applications (2025)
Wed 04/22/2026	Aragog: Just-in-Time Model Routing for Scalable Serving of Agents (2025) Cortex: Achieving Low-Latency, Cost-Efficient Remote Data Access For LLM via Semantic-Aware Knowledge Caching (NSDI ‘26)
Sun 04/26/2026	Congestion Control & Resource Management Goal: Applying classic systems concepts like congestion control and interactive debugging to the non-deterministic environment of RAG pipelines and high-volume agentic traffic. BeLLMan: Controlling Congestion in LLMs (FAISys ‘26) Murakkab: Resource-Efficient Agentic Workflow Orchestration in Cloud Platforms (2025)
Wed 04/29/2026	RAG Without the Lag: Interactive Debugging for RAG Pipelines (2025)
Sun 05/03/2026	Benchmarking & Theory Goal: Establishing rigorous, automated frameworks for measuring the reasoning capabilities, reliability, and observability of multi-agent systems in real-world scenarios AGENTSNET: Coordination and Collaborative Reasoning in Multi-Agent LLMs (2025) BLADE: Benchmarking Language Model Agents for Data-Driven Science (EMNLP ‘24)
Wed 05/06/2026	MAESTRO: Multi-Agent Evaluation Suite for Testing, Reliability, and Observability (2026) DR-Arena: an Automated Evaluation Framework for Deep Research Agents (2026)
Sun 05/10/2026	Final Presentations
Wed 05/13/2026	No class