Member of Technical Staff at Keenable.ai, building web search for AI agents. Previously Research Fellow at ielab, The University of Queensland. I build search agents, retrievers and rerankers, biomedical evidence search and efficient RAG — and release the code, models and collections behind the papers.
📈 1,522 citations · h-index 18 · i10-index 24 (Google Scholar) · 🧘 cultivation realm: Foundation Establishment (筑基期) (what?)
Most of my research code from UQ lives under @ielab — these are the repos I worked on.
| Project | What it is | |
|---|---|---|
| skim-search-agent | Composable library for building and evaluating deep-research agents over your own corpus (incl. Sieve: search, inspect, fetch) |  |
| Starbucks | 2D Matryoshka embeddings: one retriever, many sizes and depths (ECIR 2026) · 🤗 models |  |
| AutoBool | RL-trained LLM that writes Boolean queries for systematic reviews (EACL 2026) |  |
| FeB4RAG | Collection for federated search in retrieval-augmented generation (SIGIR 2024) |  |
| DiffRetriever | Diffusion language models as retrievers: parallel masked positions give dense and sparse representations |  |
| sysrev-seed-collection | Systematic review collection with seed studies (SIGIR 2022) |  |
| desksearch | Search your files by meaning, not just keywords — 100% local |  |
| MeSH Suggester | Local MeSH term suggestion server for medical literature search |  |
📰 Latest (auto-updated daily from shuaiwang.io)
- 📣 Oct 2026 — Joined Keenable.ai as Member of Technical Staff
- 📄 Oct 2026 — Do Conversational Agents Need Explicit Memory? A Controlled Comparison with Raw-History Retrieval
- 📣 Oct 2026 — New paper: Do Conversational Agents Need Explicit Memory? A Controlled Comparison with Raw-History Retrieval
- 📣 Sep 2026 — Paper Accepted at SIGIR-AP 2026
- 📄 Aug 2026 — ITER: Interaction-Aware Retrieval for Agentic Search
Ask my digital twin anything about my work (it answers from my site), or email me. Always happy to talk retrieval, RAG, Paradox games or a good xianxia novel.