About Me

I am currently an Agent Development / Evaluation intern at ByteDance, working on coding-agent evaluation, tool-use workflows, and benchmark integration. I will begin the UC San Diego CSE CS75 Project in September 2026.

I received my B.Agr. in Forestry as my first degree and a B.Eng. in Computer Science and Technology as my minor degree from Beijing Forestry University. My background combines applied machine learning, software engineering, and domain projects in forestry, remote sensing, and multimodal recognition.

Focus Areas:

• Agent Evaluation & Tool Use — Function-calling evaluation, sandboxed execution, multi-turn workflow analysis, and coding-agent behavior review.
• Applied Machine Learning — RAG prototypes, multimodal recognition, remote sensing tasks, and domain-specific QA systems.
• Software Engineering — Python-based backend tooling, Dockerized workflows, Git-based collaboration, and lightweight web demos.

📖 Education

• 2026.09 - 2027.12 | U M.S. / CSE CS75 Project
  University of California San Diego

• 2022.09 - 2026.06 | B B.Agr. in Forestry; B.Eng. in Computer Science and Technology (Minor)
  Beijing Forestry University

💻 Internship Experience

• 2025.08 - 2026.07 | ByteDance, Agent Development / Evaluation Intern
  - Worked on LLM agent evaluation workflows, including function-calling benchmarks, execution feedback, and result analysis.
  - Helped build and maintain tooling for coding-agent tasks, sandboxed runs, and multi-turn interaction records.

• 2024.10 - 2024.12 | LargeV Instrument, AI Algorithm Intern
  - Applied YOLO-based models to oral scan data and supported model tuning and data processing.
  - Integrated model-assisted annotation workflows to reduce manual labeling effort.

📝 Selected Publications

Selected peer-reviewed publications and manuscripts.

2025

🛠️ Representative Projects

• Code Agent Bot
  A coding-agent prototype with ReAct-style planning, terminal tool use, sandbox execution, and feedback-based repair.

• Medical RAG QA System
  A retrieval-augmented QA prototype with document indexing, answer grounding, and a lightweight interactive interface.

• Forestry Knowledge QA with LoRA Fine-tuning
  A domain QA project combining forestry-related data preparation, retrieval, and lightweight fine-tuning experiments.

• Multimodal Waterbird Recognition System
  An applied recognition project exploring image/audio inputs and backend service deployment.

🏆 Honors & Awards

• National First Prize, Multimedia Application Competition.
• Silver Award, ACM Programming Competition, Beijing Forestry University.

🔍 Skills

• Programming: Python, PyTorch, Docker, Git, MySQL; familiar with Java and C from coursework and projects.
• AI Systems: LLM evaluation, function-calling workflows, RAG prototypes, and benchmark integration.
• Interfaces & Tools: Streamlit, Gradio, Flask, basic web interfaces, and experiment/log analysis.