个人简介

胡一涛，天津大学智能与计算学部，教授，天津市TANKLAB实验室李克秋教授团队成员。博士毕业于美国南加州大学，导师为Ramesh Govindan教授。本科毕业于上海交通大学，导师为王新兵教授。

研究方向包括智能体系统、大语言模型（LLM）系统和 AI for Science (AI4Science)。通过与 IBM Watson 研究院、三星研究院、微软研究院等研究机构的合作，在 Nature、ASPLOS、EuroSys、SoCC、Ubicomp、INFOCOM、IWQoS、SIGCOMM、TPDS 和 TC 等高水平会议/期刊上发表论文三十余篇。研究获得NSFC等机构资助，并获得天津大学 U35 优秀青年人才奖、ACM SoCC'24 最佳论文奖等荣誉。担任 INFOCOM 和 HPCA 等会议程序委员会委员，以及 IEEE Transactions on Computers 副主编。

带领团队开发了高校问答大模型“天问”（Twen.ai）。基于RAG检索增强技术，天问在学习生活、升学就业、科研学术、奖学金申请等方面为天大师生提供咨询服务。天问于24年4月正式发布，每日服务数千咨询问题。近期，配合推动建设天津大学 AI 辅导员“小天老师”，小天老师已于25年9月全校官方上线，实现 AI 赋能学生成长。

代表论文

• [ICML 26] Mosaic: Unlocking Long-Context Inference for Diffusion LLMs via Global Memory Planning and Dynamic Peak Taming [paper][code]
• [ICML 26] KernelBand: Boosting LLM-based Kernel Optimization with a Hierarchical and Hardware-aware Multi-armed Bandit [paper]
• [EuroSys 26] PARD: Enhancing Goodput for Inference Pipeline via Proactive Request Dropping (CCF-A) [paper]
• [ASPLOS 26] PAT: Accelerating LLM Decoding via Prefix-Aware Attention with Resource Efficient Multi-Tile Kernel (CCF-A) [paper][code][slides]
• [TPDS 26] ViDA: Lossless VideoQA Acceleration via Selective Sparse Self-Speculation with Parallel Computational Load Management (CCF-A) [paper]
• [TPDS 26] Accelerating ML Inference via Opportunistic Pre-Loading on Serverless Clusters (CCF-A) [paper]
• [Nature 25] Delocalized Electrolyte Design Enables 600 Wh kg⁻¹ Lithium Metal Pouch Cells [paper]
• [TC 25] TightLLM: Maximizing Throughput for LLM Inference via Adaptive Offloading Policy (CCF-A) [paper]
• [TC 25] SLOpt: Serving Real-Time Inference Pipeline with Strict Latency Constraint (CCF-A) [paper]
• [INFOCOM 25] Harpagon: Minimizing DNN Serving Cost via Efficient Dispatching, Scheduling and Splitting (CCF-A) [paper]
• [INFOCOM 25] Lark: A Buffer-aware Building Block for Programmable Packet Scheduling in Datacenters (CCF-A) [paper]
• [SoCC 24] Pre-Warming is Not Enough: Accelerating Serverless Inference With Opportunistic Pre-Loading (CCF-B, Best Paper Award) [paper]
• [SIGCOMM 24] PPT: A Pragmatic Transport for Datacenters (CCF-A) [paper]
• [ASPLOS 24] FUYAO: DPU-enabled Direct Data Transfer for Serverless Computing (CCF-A) [paper]
• [IWQoS 23] High-throughput Sampling, Communicating and Training for Reinforcement Learning Systems (CCF-B) [paper]
• [TPDS 23] Accelerating Data Delivery of Latency-Sensitive Applications in Container Overlay Network (CCF-A) [paper]
• [SoCC 21] Scrooge: A Cost-Effective Deep Learning Inference System (CCF-B) [paper]
• [IoTDI 21] Rim: Offloading Inference to the Edge [paper]
• [Middleware 18] Olympian: Scheduling GPU Usage in a Deep Neural Network Model Serving System (CCF-B) [paper]
• [Ubicomp 16] ALPS: Accurate Landmark Positioning at City Scales (CCF-A) [paper]
• [INFOCOM 14] Critical Sensing Range for Mobile Heterogeneous Camera Sensor Networks (CCF-A) [paper]
• [Preprint 26] RAGPulse: An Open-Source RAG Workload Trace to Optimize RAG Serving Systems [paper][code][huggingface]
• [Preprint 26] ServerlessLoRA: Minimizing Latency and Cost in Serverless Inference for LoRA-Based LLMs [paper]

荣誉奖励

• 小米学者，小米公益基金会，2026
• 优秀本科毕设指导老师，天津大学，2025
• U35 优秀青年人才奖，天津大学，2025
• 最佳论文奖，SoCC，2024
• 优秀青年学术人才奖，天津大学，2024
• 启明学者，天津大学，2023
• 䇹政学者（交大首位），上海交通大学，2014
• SEIEE 毕业生代表，上海交通大学，2014