Letian Ruan

Hello, I'm Letian Ruan, a junior student at University of Michigan CSE Department, advised by Mosharaf Chowdhury in SymbioticLab . I am also a proud member of KVCache.AI working on Mooncake project from MADSys Group at THU, mentored by Teng Ma from Alibaba Cloud. Previously, I've also been pursuring my dual Bacholr's degree at Shanghai Jiao Tong University, advised by Shixuan Sun in Emerging Parallel Computing Center.

My research interests mainly lie in Machine Learning Systems, especially in designing efficient and scalable systems for GenAI. I am particularly interested in kvcache storage & transmission, multi-tenant LoRA serving and distributed architecture scheduling for AIGC and RLHF frameworks. The goal of my research is to push the boundary of how large-scale models can be designed, trained and efficiently served.

CV | Email | GitHub | LinkedIn

Aug '25	Become a member of SymbioticLab , guided by Mosharaf Chowdhury .
May '25	Joined the kvcache-ai community as an intern, mentored by Teng Ma from Alibaba Cloud.
Mar '25	Admitted to the University of Michigan, College of Engineering, Computer Science and Engineering Department.
Oct '24	Joined the Emerging Parallel Computing Center as a research assistant under supervision of Shixuan Sun

	Undergraduate Student \| University of Michigan, Ann Arbor August '25 - Present Member of SymbioticLab under CSE Department in College of Engineering Advisor: Mosharaf Chowdhury. Committed to serving Any-to-Any Multimodal LLMs with high throughput and low latency on distributed heterogeneous hardware clusters.
	Research Intern \| kvcache.ai May '25 - Present Working on Mooncake project from MADSys Group at THU. Mentor: Teng Ma. Committed to promoting KVCache-centric disaggregated architecture for LLM Serving and building more efficient and flexible data plane for RLHF training framework
	Undergraduate Student \| Shanghai Jiao Tong University September '22 - Present Member of UM-SJTU Joint Institute for dual Bachelor's degrees. Working in EPCC Lab under School of Computer Science. Advisor: Shixuan Sun . Committed to reducing latency for Multi-tenant concurrent LoRA serving and building disaggregated architecture for serverless graph processing

FaaSBoard: Efficient Graph Processing with a Disaggregated Architecture on Serverless Services

Under submission. [preprint]

We analyze the limitations of monolithic function architectures in graph processing, and motivate the design of a disaggregated serverless architecture. We introduce FaaSBoard, a graph processing system built on serverless cloud services to validate the effectiveness of the disaggregated architectural design. We conduct a comprehensive evaluation of FaaSBoard to demonstrate its superior performance and cost.

This template is a modification to Jon Barron's website.