Hi, I'm CuSO₄ Huang 👋

🎓 SE Graduate Student
🔬 Software Engineering · AI for SE · LLM Agent Testing

I'm interested in building reliable, testable, and secure AI systems, with a particular focus on LLM Agents and AI for Software Engineering (AI4SE).

My research journey started from automated software testing and fuzzing, and I'm currently exploring how traditional software engineering methodologies — testing, mutation, fault injection, coverage, and program analysis — can be adapted to evaluate and improve modern LLM-based agents.

---

🔬 Research Interests

• 🤖 LLM Agent Testing & Reliability

  • Agent behavior testing
  • Tool-use / API interaction testing
  • Multi-step trajectory analysis
  • Cross-tool constraint & data-flow validation

• 🧪 AI for Software Engineering

  • LLM-based software testing
  • Automated test generation
  • Program analysis with LLMs
  • Intelligent software engineering tools

• 🐛 Fuzzing & Automated Testing

  • Coverage-guided fuzzing
  • LLM-driven fuzz driver generation
  • API sequence generation
  • Mutation & fault injection

• 🔌 Agent Infrastructure

  • MCP (Model Context Protocol)
  • Tool-using agents
  • Agent runtime reliability
  • Stateful agent systems

---

🛠️ Tech Stack

Languages

Python · Java · C/C++ · Rust

AI & LLM

LLM Agents · RAG · Prompt Engineering · vLLM · Dify

Software Engineering

Software Testing · Fuzzing · Program Analysis · LSP · Tree-sitter

Development

Linux · Git · Docker · VS Code

---

📚 Currently Exploring

How can we systematically test increasingly autonomous LLM agents?

→ Agent behavior & trajectory testing
→ Tool/API interaction testing
→ Stateful agent reliability
→ Fault injection & mutation testing
→ Cross-tool constraint preservation
→ MCP security & robustness

---

💡 Research Philosophy

Bring the rigor of traditional software testing
into the world of increasingly autonomous AI agents.

I'm particularly interested in research that combines Software Engineering × Artificial Intelligence, turning ideas from testing, fuzzing, program analysis, and reliability engineering into practical methods for evaluating LLM-based systems.

---

📫 Let's Connect

I'm always interested in discussions about:

Agent Testing · AI4SE · LLM Agents · Software Testing · Fuzzing

Feel free to explore my repositories or reach out for research discussions and collaboration. 🚀

• 📫 Reach me: cusobob@gmail.com