CV
General Information
| Name | Hanfei Yu (余涵非) |
| hanfeiyu [at] berkeley [dot] edu | |
| Research Interests | AI Systems, LLM Serving, RL Infrastructures, Serverless Computing |
Education
-
Sep 2024 - Aug 2026 Doctor of Philosophy in Computer Engineering
Stevens Institute of Technology, Hoboken, NJ, USA
- Advisor: Prof. Hao Wang
- Dissertation: “Co-Designing Infrastructure, Runtime, and Model Serving for Efficient AI Systems”
-
June 2021 - Aug 2024 Doctor of Philosophy in Computer Science (transferred)
Louisiana State University, Baton Rouge, LA, USA
- Advisor: Prof. Hao Wang
- Transferred to Stevens Institute of Technology in 2024
-
Sep 2019 - Feb 2021 Master in Computer Science and Systems
University of Washington Tacoma, Tacoma, WA, USA
- Advisor: Prof. Wes J. Lloyd
- Thesis: “FaaSRank: Learning to Schedule Functions in Serverless Platforms”
-
Sep 2015 - July 2019 Bachelor in Electronic Engineering
Shanghai Jiao Tong University, Shanghai, China
Experience
-
Sep 2026 - Present Postdoctoral Scholar
Sky Computing Lab, UC Berkeley, Berkeley, CA, USA
Working with Prof. Ion Stoica on efficient AI systems and LLM infrastructure
-
June 2026 - Aug 2026 Student Researcher at Seed Training Infra Team
ByteDance Seed, Bellevue, WA, USA
- Optimized sandbox snapshot techniques for efficient agent serving and replay
- Designed and built production data feedback pipelines for evaluating Seed models on agentic rollouts
-
May 2025 - Aug 2025 Research Intern at Efficient AI Group
Microsoft 365 Research, Redmond, WA, USA
- Characterized production workloads of OpenAI Copilot services
- Designed and implemented efficient KV cache management solutions for multimodal LLM serving
-
May 2024 - Aug 2024 Research Intern at Azure Research Systems
Microsoft Azure Research, Redmond, WA, USA
- Characterized production workloads of Azure container platforms
- Designed new solutions to optimize the resource efficiency of container orchestration
-
Aug 2018 - Feb 2019 Software Developer Intern at UEFI-BIOS Development Group
Intel, Shanghai, China
- Developed a cross-platform UEFI driver that analyzes network traffic with PCAP
Awards
-
From 2021 Student Travel Scholarships
AAAI'24, WWW'22, ACSOS'21