Zhiyu Yang · CS PhD @ UT Dallas

I study how collaborative AI agents improve over time.

Portrait of Zhiyu Yang

AI agents, real-world evaluation, and human–AI collaboration.

Agents that persist, learn, and adapt.

The common thread is continuity: how an agent keeps its bearings, learns from experience, and adjusts as the task, context, and relationship with a person evolve.

Persistent agents

Maintaining context, state, and task continuity across long-running human–agent work.

Learning from interaction

Turning real feedback, corrections, and outcomes into signals that can improve future behavior.

Adaptive agent systems

Helping agents adjust their behavior and effort as tasks, feedback, and collaboration with people evolve.

Selected publications.

All publications

DSDBench

EMNLP 2025 Main · Oral

Why Stop at One Error? Benchmarking LLMs as Data Science Code Debuggers for Multi-Hop and Multi-Bug Errors

A data-science debugging benchmark for tracing multi-hop logical errors and resolving multiple bugs in realistic code.

MatPlotAgent

ACL 2024 Findings

Method and Evaluation for LLM-Based Agentic Scientific Data Visualization

A visual-feedback agent that turns natural-language requests into scientific visualizations, then uses the rendered chart to repair its own code.

CompactRAG

WWW 2026

Reducing LLM Calls and Token Overhead in Multi-Hop Question Answering

A cost-efficient multi-hop reasoning framework that separates offline corpus restructuring from online inference and keeps the main LLM to two calls.

Academic and research background.

Education and research experience.

University of Texas at Dallas

CS PhD

Singapore Management University

Research Assistant

Mentored by Yang Deng

THUNLP · ModelBest · OpenBMB

Research Intern

Mentored by Shuo Wang

Beijing Language and Culture University

M.Eng. in Computer Science and Technology

Sichuan University

B.Eng. in Computer Science and Technology