Hi! Welcome to my website.

My name is Zikai Wang. I am a 2nd year PhD student in Khoury College of Computer Sciences at Northeastern University, advised by Prof. Cheng Tan. I am a member of the System Research Group, where I work on building reliable and verifiable data systems.

Before Northeastern, I obtained my M.E and B.E in Renmin University of China (RUC) under the supervision of Prof. Yunpeng Chai. Where I worked on the distributed database system.

Education

• PhD in Computer Science, Northeastern University
  • Sep 2024 - present
  • Advisor: Prof. Cheng Tan.
• Master of Engineering in Computer Science, Renmin University of China
  • Sep 2021 - Jun 2024
  • Advisor: Prof. Yunpeng Chai
• Bachelor of Engineering in Computer Science, Renmin University of China
  • Sep 2017 - Jun 2021

Research Experience

• Graduate Research Assistant, Northeastern University
  • Sep 2024 - present
  • Advisor: Prof. Cheng Tan.
  • Working on reliable and verifiable data systems
• Research Intern, Microsoft Research Asia
  • May 2024 - Aug 2024
  • Advisor: Qianxi Zhang, Baotong Lu
  • Working on the robustness of vector index. We proposed a metric robustness to measure the recall distribution of vector indices, and provided index selection and tuning guidelines for applications relying on vector indices like Retrieval-Augmented Generation based on the robusntess metric.
• Research Intern, TU Darmstadt
  • Mar 2023 - May 2024
  • Advisor: Prof. Carsten Binnig, Dr. Tobias Ziegler
  • Working on the scalable transactional database system above disaggregated shared-caching system.
• Research Intern, Tencent
  • Sep 2020 - Sep 2022
  • Working on the distributed database system. I provided heterogeneous multiple replicas supports for TDSQL3, a commercial distributed HTAP database in C++ developed by Tencent. I proposed a workload-aware storage model based on the key-value storage engine of TDSQL3.
• Research Intern, Renmin University of China
  • Sep 2019 - Sep 2020
  • Advisor: Prof. Yunpeng Chai, Dr. Yangyang Wang
  • Working on optimizing the distributed consensus protocol for distributed key-value stores.