Kaituo (Philip) Zhang · 章凯拓

Building AI that deserves to be trusted.

Hello, my name is Kaituo (Philip) Zhang. I’m a PhD candidate in the Department of Industrial Engineering at the University of Houston, co-advised by Dr. Na Zou and Dr. Ying Lin. My current research interests include fairness and faithfulness in large language models (LLMs) and context management.

Trustworthy LLMsSynthetic DataAnomaly Detection
Kaituo Zhang looking toward a sunset
PhD Candidate · Research AssistantHouston, United States

The LORD is my shepherd; I have all that I need.

Psalm 23:1 · NLT
01

News

  1. AsynCodeBench is available on arXiv.Paper ↗Code ↗
  2. SRD was accepted to Findings of EMNLP 2026.Code ↗Paper ↗
  3. Our survey on evaluating LLM-generated synthetic data was accepted by TMLR.Repo ↗Paper ↗
  4. Tool-use Tax is available on arXiv.Paper ↗
  5. I passed the screening exam and became a PhD candidate.
  6. Our anomaly detection work was accepted by Expert Systems with Applications.Code ↗Paper ↗
02

Selected work

01

AsynCodeBench

A dependency-centric benchmark for evaluating collaboration in asynchronous multi-agent software engineering.

Multi-Agent SystemsSoftware Engineering
02

EDROD

A parameter-free, unsupervised anomaly detection algorithm based on entropy density ratios.

Anomaly DetectionResearch
03

SRD

A self-reflective detoxification framework for building safer and more trustworthy large language models.

Trustworthy LLMsDetoxification