Luo Mai

PIA-Bench: Towards Automated Privacy Impact Assessment with Large Language Models

Jiamin Zheng, Hao-Ping Lee, Luo Mai, Jingjie Li

Privacy impact assessment (PIA) is a critical instrument for institutions to proactively identify privacy risks and develop mitigation strategies before system deployment. While mandated across regulatory and institutional contexts, executing PIA requires extensive privacy and technical expertise, posing a particular challenge for …

Rethinking AI Cloud Infrastructure for Agentic Serving Systems with the Aries Experimentation Framework

Leonid Kondrashov, Hongrui Liu, JooYoung Park, Boxi Zhou et autres

Autonomous agents challenge conventional LLM serving by coupling repeated inference with persistent context and sandboxed tool execution. We present Aries, a full-stack experimentation framework that separates task semantics from execution configurations, reconstructs cross-component agent trajectories with correlated system telemetry, and exposes stateful …

Rethinking AI Cloud Infrastructure for Agentic Serving Systems with the Aries Experimentation Framework

Leonid Kondrashov, Hongrui Liu, JooYoung Park, Boxi Zhou et autres

Autonomous agents challenge conventional LLM serving by coupling repeated inference with persistent context and sandboxed tool execution. We present Aries, a full-stack experimentation framework that separates task semantics from execution configurations, reconstructs cross-component agent trajectories with correlated system telemetry, and exposes stateful …

us, sg, gb (code pays fourni par la source)

TileSight: A First-Principles Tile-Centric Analytical GPU Performance Model from Cores to Clusters

Zhiwen Mo, Y Cheng, Lei Wang, Tang Z et autres

Recent GPU programming frameworks such as Triton, TileLang, and CUDA Tile adopt tiles as first-class primitives, making tile-centric programming the prevailing approach for high-performance GPU kernels. Performance-analysis tooling has not followed: programmers still rely on coarse roofline bounds, opaque ML predictors, or …

TileSight: A First-Principles Tile-Centric Analytical GPU Performance Model from Cores to Clusters

Zhiwen Mo, Yu Cheng, Lei Wang, Zhengju Tang et autres

Recent GPU programming frameworks such as Triton, TileLang, and CUDA Tile adopt tiles as first-class primitives, making tile-centric programming the prevailing approach for high-performance GPU kernels. Performance-analysis tooling has not followed: programmers still rely on coarse roofline bounds, opaque ML predictors, or …

gb, us, cn (code pays fourni par la source)

BatchGen: An Architecture for Scalable and Efficient Batch Inference

Tairan Xu, Leyang Xue, Zhan Lu, Jinfu Deng et autres

Batch inference has become a central mode of AI computation, yet existing inference engines still rely on execution models designed for interactive serving. When scaled to millions of sequences, batch workloads reveal two fundamental requirements: the ability to handle extreme inter- and …

SwarmX: Agentic Scheduling for Low-Latency Agentic Systems

Yeqi Huang, Yanwei Ye, Guomin Chen, Wenhao Su et autres

Agentic AI applications compose multiple model calls and tool executions, creating new scheduling challenges for GPU-CPU clusters. Their inference time and model-call structure often depend on prompt semantics, making conventional scheduling approaches ineffective for low-latency serving. This paper presents SwarmX, a system …

BatchGen: An Architecture for Scalable and Efficient Batch Inference

Tairan Xu, Leyang Xue, Zhan Lu, Jinfu Deng et autres

Batch inference has become a central mode of AI computation, yet existing inference engines still rely on execution models designed for interactive serving. When scaled to millions of sequences, batch workloads reveal two fundamental requirements: the ability to handle extreme inter- and …

cn (code pays fourni par la source)

SwarmX: Agentic Scheduling for Low-Latency Agentic Systems

Yeqi Huang, Yanwei Ye, Guomin Chen, Wenhao Su et autres

Agentic AI applications compose multiple model calls and tool executions, creating new scheduling challenges for GPU-CPU clusters. Their inference time and model-call structure often depend on prompt semantics, making conventional scheduling approaches ineffective for low-latency serving. This paper presents SwarmX, a system …

gb, cn (code pays fourni par la source)

Ryze: Evidence-Enriched Data Synthesis from Biomedical Papers

Yeqi Huang, Yue Chen, Yanwei Ye, Guanhao Su et autres

General-purpose VLMs remain unreliable for biomedical research because valid answers in scientific papers depend on evidence split across figures, tables, charts, captions, and referring text. Existing post-training pipelines are bottlenecked by costly expert annotation and by synthetic data that drops this evidence …

Ryze: Evidence-Enriched Data Synthesis from Biomedical Papers

Yeqi Huang, Yue Chen, Yanwei Ye, Guanhao Su et autres

General-purpose VLMs remain unreliable for biomedical research because valid answers in scientific papers depend on evidence split across figures, tables, charts, captions, and referring text. Existing post-training pipelines are bottlenecked by costly expert annotation and by synthetic data that drops this evidence …

gb (code pays fourni par la source)

Specialize Roles, Mix Deployments: Pushing the Cost-Accuracy Frontier of LLM Agent Teams

Yu Jiang, Liang Cheng, Yeqi Huang, Yufan Zhao et autres

LLM agents are increasingly deployed as multi-role teams, where tasks are divided across specialized roles such as planner, executor, and verifier. In these systems, cost and accuracy are no longer properties of a single model: they depend on which model fills each …