CV · Zihang Zhou

Education

Shanghai Jiao Tong University (SJTU)

Shanghai, China

School of Artificial IntelligenceSep 2024 – Present

B.Eng. Candidate in Artificial Intelligence

GPA: 4.04 / 4.3 (Average Score: 92.7)
Honors: National Scholarship (Top 0.5%, Oct 2025);
Zhiyuan Honorary Scholarship.
Early Admission: Admitted one year early (Grade 11) via the National College Entrance Exam (Gaokao); ranked in the top 0.05% in Jiangsu Province.

Research Interests: Self-evolving LLM agents, tool learning, agent evaluation, and reliable and reproducible agentic systems.

Publications & Preprints

Agentic Time Machine as an Infrastructure for Future-Event Forecasting.
EMNLP 2026 Main Conference (Oral).
SetupX: Can LLM Agents Learn from Past Failures in Functionality-Correct Code Repository Setup?.
First Author. Under review at ICLR 2027.
Workspace-Bench 1.0: Benchmarking AI Agents on Workspace Tasks with Large-Scale File Dependencies.
NeurIPS 2026 (Poster).
LLM/Agent-as-Data-Analyst: A Survey.
Under review at IEEE TKDE.

Internship Experience

Jun 2026 – Present

Microsoft Research Asia (MSRA)

Shanghai, China

Research Intern

Project: Enterprise Workflow Benchmark for LLMs and Agents
Designed enterprise workflow tasks and constructed evaluation datasets to assess LLM and agent capabilities in enterprise scenarios.
Defined evaluation metrics and developed the evaluation framework for systematic assessment of model and agent performance on workflow tasks.

Research Experience

Apr 2026 – Jun 2026

Multi-Agent Governance & Intelligence Crew (MAGIC)

Shanghai, China

Undergraduate Researcher

Project: Agentic Time Machine
Paper: Agentic Time Machine as an Infrastructure for Future-Event Forecasting. EMNLP 2026 Main Conference (Oral).
Evaluated the performance of agents equipped with the Time Machine infrastructure on future-event forecasting tasks.

Oct 2025 – Apr 2026

SJTU Data Analysis Study Group

Shanghai, China

Undergraduate Researcher

Project: SetupX – Autonomous Repository Setup
First-author paper: SetupX: Can LLM Agents Learn from Past Failures in Functionality-Correct Code Repository Setup?. Under review at ICLR 2027.
Developed an autonomous agent for complex GitHub repository setup, integrating an external knowledge base via RAG to help resolve dependency conflicts and configuration errors.
Refined the agent–environment interaction loop to improve automated repository setup success rates.

Jun 2025 – Oct 2025

SJTU Data Analysis Study Group

Shanghai, China

Undergraduate Researcher

Project: LLM/Agent-as-Data-Analyst Survey
Co-authored LLM/Agent-as-Data-Analyst: A Survey. Under review at IEEE TKDE.
Authored the sections on Video and 3D Models, systematically reviewing state-of-the-art methods for multimodal data analysis agents.
Curated and maintained the awesome-data-llm repository to track the latest research.

Skills

Programming:: Python, C++
Libraries & Tools:: PyTorch, NumPy, pandas, Matplotlib, LangChain, Docker, Git, LaTeX
Machine Learning:: Neural network architectures, model training, evaluation, and experimental design
LLMs & Agents:: Tool use, retrieval-augmented generation, multi-agent systems, agent evaluation, and repository automation

Relevant Coursework

Mathematics:

Mathematical Analysis, Linear Algebra, Numerical Analysis, Probability and Statistics, Discrete Mathematics

Computer Science:

Data Structures, Algorithm Design and Analysis, C++ Programming, Machine Learning, Deep Learning, Reinforcement learning, Introduction to Intelligent Robotics, Computer Architecture