Portrait of Wenxuan Xia

Wenxuan Xia

PhD student in Data Science and Analytics

I am a PhD student at The Hong Kong University of Science and Technology (Guangzhou). My research focuses on efficient vector search and data systems, particularly approximate nearest-neighbor search, filtered search, and graph-based vector indexes.

Outside research, I enjoy building personal agent systems and small tools for learning, research, and everyday workflows.

Education

The Hong Kong University of Science and Technology (Guangzhou)

PhD in Data Science and Analytics, September 2024 to present

Advisor: Wei Wang

Shenzhen University

BEng in Software Engineering, September 2019 to July 2023

Undergraduate thesis advisor: Jianbin Qin

GPA: 4.21/4.5; ranked first in the major.

Publications

† equal contribution (co-first authors) · * corresponding author

E2E: Efficient Filtered AKNN Search via Adaptive Termination

Wenxuan Xia†, Mingyu Yang†*, Wentao Li, Wei Wang*.

ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 2026. DOI Paper Code

Elastic Index Selection for Label-Hybrid AKNN Search

Mingyu Yang, Wenxuan Xia, Wentao Li, Raymond Chi-Wing Wong, Wei Wang*.

Proceedings of the VLDB Endowment, 19(4): 617–630, VLDB 2026. DOI Paper Code

HRNN: A Hybrid Graph Index for Approximate Reverse k-Nearest Neighbor Search on High-Dimensional Vectors

Wenxuan Xia, Mingyu Yang, Wentao Li, Wei Wang.

arXiv preprint, 2026. Paper Code

Selected projects

My private NanoClaw

I maintain a private fork of NanoClaw as a personal AI assistant, extensively customizing its runtime and building new tools to support my daily routines and learning. I also use this project to explore software development in the agentic era, using Claude Code, Codex, and agent teams to build and maintain a continuously running agent system.

Local weather map showing conditions around Pingshan and nearby towns.
Local weather at a glance.
Art-learning conversation with follow-up lesson controls and an illustrated museum-viewing guide.
Learning about art through conversation and visual guides.
NanoClaw course dashboard showing ongoing courses, learning progress, and activity.
Managing ongoing courses and tracking learning progress.

Paper posters

I am fascinated by typography. I am not a professional designer, but I am honored to have had the chance to design the posters for our papers.

Poster for the KDD 2026 paper E2E, a dense landscape layout with problem definition, motivation, the early-probe-to-adaptive-termination method, and experimental results.
Poster for E2E at KDD 2026 — 100 × 85 cm, set in Avenir Next with Avenir Next Condensed display. Click the image for the full-resolution PDF.
Poster for the VLDB 2026 paper Elastic Index Selection, an A1 landscape layout in a plum palette with numbered sections from problem to takeaway.
Poster for Elastic Index Selection at VLDB 2026 — A1 landscape, set in Myriad Pro display over a DIN Pro body. Click the image for the full-resolution PDF.

Teaching

Teaching assistant at HKUST(GZ):

Honors & awards