Xueqiang (Patrick) Xu

I am a first-year CS PhD student at UIUC, where I am advised by Prof. Jiawei Han. My research is on building trustworthy LLM agent systems — agents that keep learning from memory and external knowledge, and that stay honest, harmless, and steerable as they grow more capable. I completed my undergraduate studies at UIUC, graduating with a Highest Honors B.S. in Computer Science. I work with Shi Feng this summer on LLM safety alignment.

徐学强  /  Email  /  CV  /  Google Scholar  /  GitHub  /  LinkedIn

Photo by Ben Wu

[Jun 2026] 🚀 New preprint: Harness-1, a 20B search agent trained with RL inside a state-externalizing harness — it matches frontier-model searchers on agentic search while staying fully open-source (code).
[May 2026] 🎉 Two papers accepted: one on LLM Reranking at ICML 2026, and one on Compact LLM Reranking at KDD 2026.
[Jan 2026] One paper on zero-shot entity structure extraction ZOES has been accepted by EACL 2026 Main Conference.

Preprint DOI Code Poster

Instead of full text, CoRank reranks compact LLM-extracted document features — category, section, keywords — so roughly 10× more candidates fit in one prompt. Training-free, it lifts average nDCG@10 from 50.6 to 55.5 at 40% of the token cost.

Preprint

ZOES extracts entity structures with no schema and no annotated samples, through a cycle of enrichment, refinement, and unification — an entity and its structure reinforce each other.

Preprint Code

s3 is a lightweight, model-agnostic framework that decouples the searcher from the generator, trained with RL on only 2.4k examples.

arXiv Code

A survey that unifies agentic AI research into one framework spanning agent adaptation and tool adaptation — how foundation models are adapted to plan, reason, and use external tools.

Preprint Code

TELEClass pairs the general knowledge of LLMs with features mined from an unlabeled corpus, enriching the label taxonomy for hierarchical text classification under minimal supervision.

• City Scholar at UIUC
• Illinois Scholars Undergraduate Research
• IIDAI scholar

Away from research I like keeping a record of my life, so a camera usually comes along — cities, cats, and whatever the light happens to be doing that day.

Template from Jon Barron