CV

Education

• Ph.D. Candidate, University of Science and Technology of China, Institute of Advanced Technology and Future Network Laboratory, 2023.09 - Present
• Bachelor’s Degree, Chongqing University of Posts and Telecommunications, School of Communication and Information Engineering, 2019.09 - 2023.06

Research interests

• Efficient LLM inference serving and deployment
• AI infrastructure for heterogeneous GPU clusters
• Disaggregated serving, scheduling, and resource orchestration
• Multi-agent systems and communication-efficient collaboration
• Multimodal model efficiency and visual token pruning

Research experience

DisHelis: Serving Disaggregated Large Language Models over Heterogeneous Environments via Hierarchical Max-Flow

• Studied resource allocation and model deployment for disaggregated LLM serving on heterogeneous GPU clusters.
• Modeled heterogeneous GPU compute, network bandwidth, and stage-level heterogeneity in disaggregated inference.
• Proposed a hierarchical max-flow-graph-based deployment algorithm and a lightweight redeployment mechanism for fluctuating online conditions.
• Evaluated against DistServe, HexGen2, Helix, and vLLM-style baselines.

FAESR: Fine-Grained Rate Adaptation for Energy-Aware Super Resolution in Mobile Panoramic Video Streaming

• Proposed a fine-grained bitrate adaptation method for neural-network-enhanced panoramic video streaming.
• Built a super-resolution power model from real-world measurements and formulated a joint QoE and energy-aware optimization problem.
• Developed a branch-sequence deep reinforcement learning algorithm for tile-level bitrate allocation.
• Publication: IEEE Transactions on Cognitive Communications and Networking, Early Access 2025, first author.

Honors

• National Scholarship, University of Science and Technology of China, 2025
• Graduate Academic First-Class Scholarship, University of Science and Technology of China, 2023 and 2024
• MathorCup National Undergraduate Mathematical Modeling Competition, National First Prize, 2022
• Outstanding Graduate of Chongqing Municipality, 2023
• Minister of Mobile Development Department, Hongyan Online School Workstation; Class Monitor, Excellent Engineer Class, 2020 - 2023

Skills

• Research: literature search with CNKI, IEEE Xplore, Google Scholar, GitHub, and related research resources
• Systems: LLM inference serving, scheduling, distributed databases, backend systems, traffic data processing
• Development: Web frontend/backend development and Android mobile application development
• Language: English reading and writing for professional literature; CET-4 and CET-6

Publications

LatCom: Latent Compression for Efficient Multi-Agent Collaboration

Co-first author · EMNLP 2026 · Poster · 2026

GSTEP: Global Spatio-Temporal Density-Driven Visual Token Pruning for Efficient Video Large Language Models

Co-first author · ACM Multimedia 2026 · Poster · 2026

SAVP: Scene-Aware Vision Token Pruning for Efficient Video Large Language Models

Co-first author · EMNLP 2026 · Poster · 2026

SpecCache: Speculative KV Cache Reuse for Efficient RAG Serving

Co-first author · ACL 2026 · Oral · 2026

HAWK: Head Importance-Aware Visual Token Pruning in Multimodal Models

Co-first author · CVPR 2026 · Poster · 2026

Multi-Timescale Joint Optimization of Task Scheduling, Instance Switching, and Resource Scaling for Disaggregated LLM Serving

First author · IEEE Transactions on Cognitive Communications and Networking · SCI 二区 · 2026

DisHelis: Optimizing Deployment of Disaggregated LLMs Inference Serving over Heterogeneous Environments via Hierarchical Max-Flow

First author · IEEE Transactions on Cognitive Communications and Networking · SCI 一区 · 2026

FAESR: Fine-Grained Rate Adaptation for Energy-Aware Super Resolution in Mobile Panoramic Video Streaming

First author · IEEE Transactions on Cognitive Communications and Networking · SCI 一区 · 2025