Meituan · Multimodal LLM Inference Acceleration & Long-horizon AI Agent
Inference acceleration for Qwen3-VL / Qwen3.5 and later multimodal LLMs; cost reduction and efficiency optimization of long-horizon AI Agents.
01
About Me
I am Zhangkunlong, an engineer focused on shipping AI systems to production, based in Shanghai. I currently work at Meituan on multimodal LLM inference acceleration and long-horizon AI Agent optimization.
I graduated from Southern University of Science and Technology (B.Eng. in Computer Science and Technology, M.Eng. in Electronic Science and Technology). I have contributed pull requests to the vLLM community, published a paper on hardware-aware DNN compression for edge devices (Best Poster Award), and won several competitions on Tianchi and Kaggle.
02
Experience
2026.07 - Present
2025.04 - 2025.09
Meituan · Multimodal LLM Inference Acceleration (Intern)
Contributed to inference acceleration for the Qwen2.5-VL multimodal model.
03
Education
Southern University of Science and Technology
Electronic Science and Technology
Southern University of Science and Technology
Computer Science and Technology
04
Skills
LLM Inference Acceleration
Long-horizon Agent Optimization
Multimodal Models
Languages & Tools
05
Publications
Best Poster Award
Zhang K, Li G, Lu N, et al. — 2025 7th International Conference on Data-driven Optimization of Complex Systems (DOCS), IEEE, 2025, pp. 387–392.
06
Competitions
Tianchi · 3rd Place
Fund trend simulation prediction track; 3rd place out of 813 teams.
Want to see what I am working on?
Notes on LLM inference acceleration, AI Agents, and engineering practice.