Personal introduction
About me
I am an M.S. student at the National University of Defense Technology. My interests include using LLMs more efficient and applying multi-agent architectures to practical problem solving.
My current research focuses on AIOps — using data-driven methods and AI to automate IT operations such as log analysis and incident management. I look forward to connecting and discussing related topics.
Education
2019.6-2023.6 B.S. in Software Engineering, School of Computer Science, National University of Defense Technology
2023.9–2026.6 M.S. in Software Engineering, School of Computer Science, National University of Defense Technology
Lab & Supervisor
At present, I am studying in the State key laboratory of parallel and distributed processing(PDL), and the supervisor is Academician Huaimin Wang. At the same time, Professor Dawei Feng provides specific guidance with great patience ♥.
Publications
- A Survey on the Application of Large Language Model-Based Agents in Root Cause Analysis of Software Systems — Submitted
Internships
Tencent, CSIG — SRE Intern (2025.07–2025.09)
- Built LLM- and statistics-driven alert triage to suppress noise and prioritize business-impacting incidents.
- Automated gateway alert clustering and report generation via NLP and aggregation analysis, improving analysis efficiency and decision support.
- Impact: Daily alerts reduced from 665 to 160 (−76%); alert categories reduced from 1,823 to 428 (−77%).
iFLYTEK, AI Engineering Institute — Assistant AI R&D Engineer (2024.08–2024.09)
- Fine-tuned Spark 13B with LoRA for spike appid prediction from 3,600+ sample telemetry, achieving 87% accuracy on surge scenarios.
- Implemented multi-agent workflow for end-to-end crash localization via logs, stack traces, and report generation.
- Built RAG pipeline on 13.78M XFS log lines with vectorized expert playbooks to auto-generate error analysis reports.