Talks & Seminars

The Wrong Contract at Every Layer: Redesigning OS and Hardware for AI Datacenters

May 15, 2026 Dimitrios Skarlatos — Carnegie Mellon University

Scaling Distributed Systems Designs

May 8, 2026 Jialin Li — National University of Singapore

Managing and Securing Google's Fleet of Multi-Node Servers

May 8, 2026 Jeff Mogul — Google

System Architectures for serving Agentic applications

May 1, 2026 Saurabh Agarwal — UT Austin

Understanding how agents are deployed in production

April 22, 2026 Melissa Pan — UC Berkeley

Rethinking Database Optimization for Modern Workloads

April 8, 2026 Audrey Cheng — UC Berkeley

Evaluation beyond benchmarks: Understanding what models do, why they fail, and what users want

March 6, 2026 Lisa Dunlap — UC Berkeley

RLVE: Scaling Up Reinforcement Learning for Language Models with Adaptive Verifiable Environments

February 6, 2026 Zhiyuan Zeng — University of Washington

Rethinking LLM Serving From the Application’s Perspective

December 12, 2025 In Gim — Yale University

Effectively Scaling Reinforcement Learning for LLMs

December 1, 2025 Yi Wu — Tsinghua University

RDMA P2P Communication Patterns for KvCache Transfer, Weight Update, and MoE Routing

November 21, 2025 Lequn Chen — Perplexity

GEPA and prompt optimization for compound AI systems

November 7, 2025 Lakshya Agrawal — UC Berkeley

Rethinking Prediction for System Tuning and Architectural Modeling

October 17, 2025 Jonathan Balkind — UCSB

SOSP practice talk: Yicheng Liu (UCLA)

October 10, 2025 Yicheng Liu — UCLA

MoDM: Mixture of Diffusion Models to Navigate the Performance-Quality Trade-off in Image Generation

October 3, 2025 Nishil Talati — UIUC

MSCCL++: Rethinking GPU Communication Abstractions for AI

September 12, 2025 Aashaka Shah & Roshan Dathathri — Microsoft

Acceleration of diffusion models

September 5, 2025 Xingyang Li — MIT

Two fronts of AI: addressing performance and power/cooling challenges

August 22, 2025 Esha Choukse — Microsoft

Edge data center: experience and lessons in building the world's largest cluster of edge devices

August 15, 2025 Animesh Dangwal — UCSB

Whole-Body Conditioned Egocentric Video Prediction

August 1, 2025 Yutong Bai — UC Berkeley

Algorithm–system co-design of late binding techniques for adaptive serving

July 18, 2025 Yeonju Ro — University of Texas Austin

LiquidCache, a novel pushdown-based disaggregated caching system

July 11, 2025 Xiangpeng Hao — University of Wisconsin Madison

Training with Confidence: Catching Silent Errors in Deep Learning Training with Automated Proactive Checks

June 27, 2025 Yuxuan Jiang — University of Michigan

EvalTree: Profiling Language Model Weaknesses via Hierarchical Capability Trees

June 20, 2025 Zhiyuan Zeng — University of Washington