CV
This is my Resume.
General Information
| Full Name | Zihe Song (宋子鹤) |
| zihe.song@utdallas.edu | |
| Website | https://zihesong.github.io |
| Location | Richardson, Texas, USA |
| Summary | PhD candidate in Computer Science at UT Dallas. Research focuses on LLM-driven UI agents, mobile testing, software reliability, and efficiency robustness of deep models. |
Education
-
2020–present PhD in Computer Science
The University of Texas at Dallas - {"Advisor"=>"Dr. Wei Yang."}
- Specializing in Android systems, program analysis, ML infrastructure, and large-scale debugging and testing.
- Publications across ISSTA, USENIX Security, CVPR, ASPLOS, ICSE, MobileSoft.
- Finalist in Amazon Nova AI Challenge.
-
2018–2020 M.S. in Computer Science
The University of Texas at Dallas -
2014–2018 B.Eng. in Communication Engineering
University of Electronic Science and Technology of China (UESTC)
Competitions
-
2025 Amazon Nova AI Challenge – Finalist
Amazon Science - Published in Amazon Science Trusted AI 2025.
- Designed adversarial attack strategies uncovering high-severity LLM vulnerabilities.
- Contributed to COMET, improving evaluation throughput and automated red-teaming.
Publications
- TaOPT: Tool-agnostic optimization of parallelized automated mobile UI testing, ASPLOS 2025
- Can You Mimic Me? Exploring the use of Android Record & Replay tools in debugging, MOBILESoft 2025
- An investigation on numerical bugs in GPU programs towards automated bug detection, ISSTA 2025
- SoK: Efficiency robustness of dynamic deep learning systems, USENIX Security 2025
- Guardian: A runtime framework for LLM-based UI exploration, ISSTA 2024
- WEFix: Intelligent automatic generation of explicit waits for efficient web end-to-end flaky tests, WWW 2024
- NICGSlowDown: Evaluating the efficiency robustness of neural image caption generation models, CVPR 2022
- An empirical analysis of compatibility issues for industrial mobile games, ISSRE 2022
- NMTSloth: Understanding and testing efficiency degradation of neural machine translation systems, ESEC/FSE 2022
- An empirical analysis of UI-based flaky tests, ICSE 2021
- An automated framework for gaming platform to test multiple games, ICSE SRC 2020
Projects
-
Multimodal Agent Framework for Android App Generation (ICLR 2026 under review)
- Built a benchmark of 101 Android tasks requiring LLMs to generate full apps.
- Built a multi-agent system for documentation summarization, UI navigation, and test-case generation.
- Built an evaluation pipeline showing current LLMs reach ~18.8% functional correctness.
-
LLM-Driven Smartphone Interaction Enhancement
- Built an LLM-driven framework improving intent disambiguation and UI navigation robustness.
- Developed multi-agent reasoning using screenshots, view hierarchies, and natural-language instructions.
- Designed clarification-prompt models improving success rates for UI navigation tasks.
Internship
-
2020 Research Intern
NetEase Fuxi AI Lab (Hangzhou, remote) - Built automated gameplay testing systems for diverse player behavior simulation.
- Applied GAIL-based RL for trajectory generation reducing QA workload.
- Conducted large-scale compatibility study informing debugging workflows and device optimization.
Research Supervision
- Supervised master's students on software engineering research; built multilingual automated bug-repair data collection framework; paper submitted to TOSEM.
Teaching Experience
- CS 1325 – Introduction to Programming (C/C++), Instructor, UT Dallas, Spring 2026.
- ECS 1100 – Introduction to Engineering and Computer Science, Instructor, UT Dallas, Fall 2025.
- Teaching Assistant for AI, Algorithm Design & Analysis, Big Data Analytics (2020–2025).
Technical Skills
-
Programming
- Python
- Java
- Kotlin
- C/C++
- Bash
-
Machine Learning / AI
- LLMs
- Generative AI
- NLP
- Reinforcement Learning
- ML Evaluation
- Adversarial Testing
-
Systems & Mobile
- Android Framework
- ADB
- UIAutomator
- View Hierarchy Analysis
- Profiling
- Tracing
-
Program Analysis
- Static Analysis
- Dynamic Analysis
- Debugging Tools
- Runtime Monitoring
-
Tools & Frameworks
- CI/CD
- Test Automation
- Profiling Pipelines
- PyTorch
- TensorFlow
- Docker
- Git
- Linux
Languages
- Chinese (Native)
- English (Fluent – research & teaching)
Interests
- LLM agents, mobile UI testing, software reliability
- Tennis, skiing, baking, travel, photography