CV
Contact
- Email: moon212549@gmail.com
- Location: Guangzhou, China
- GitHub: PoetMoon
Education
South China University of Technology
Bachelor of Computer Science
GPA: 3.86 / 4.00, Rank: 13 / 157
Main courses: Linear Algebra (98), Algorithms (98), Compiler Principles (97), Computer Architecture (97), Software Engineering (96), C++ (95), Probability Theory and Mathematical Statistics (95)
Research Experience
The Hong Kong University of Science and Technology (Guangzhou)
Research Assistant, supervised by Prof. Chengwei Qin
University of Cambridge
Visiting Student, August 2025
Publication
Y. Wu, Y.-T. Zhong, Z.-Y. Ma, & Y.-J. Gong. (2026). GeM-EA: A Generative and Meta-learning Enhanced Evolutionary Algorithm for Streaming Data-Driven Optimization. GECCO 2026.
Research / Project Experience
LLM & Genetic Programming for RPA Automation
Core Member, May 2024 - Apr 2025
- Applied Parameter-Efficient Fine-Tuning (LoRA) to adapt Large Language Models for specific business domains.
- Combined LLMs with Genetic Programming to generate and optimize complex workflow structures with multi-level branches.
- Achieved a 15% increase in execution efficiency and a 23% boost in complex logic handling capability.
Intelligent Vehicle Navigation System
Leader, Feb 2025 - May 2025
- Implemented Delaunay triangulation and KD-Tree for spatial indexing and road network construction.
- Developed path planning strategies using A* for static shortest paths and Dijkstra with BPR functions for dynamic traffic simulation.
Data-Driven End-to-End Automatic Black-Box Optimization Framework
Core Member, Problem Modeling Group, Jan 2025 - Feb 2026
- Engineered an automated framework integrating Deep Reinforcement Learning (PPO) with a heterogeneous surrogate model pool, including Gaussian Process and RBFN models.
- Devised a generative replay and bi-level meta-learning mechanism to address concept drift in streaming data.
- Implemented PPO-based dynamic variable grouping to mitigate high-dimensional optimization challenges.
- Designed a reinforcement-learning state space for adaptive mutation and crossover operator selection.
- Achieved strong results on SDDObench and CEC2013 LSGO, and applied the framework to day-ahead power grid scheduling.
Intelligent Recipe Q&A System Based on RAG Architecture
Core Developer, Feb 2026 - Mar 2026
- Designed and developed an end-to-end Retrieval-Augmented Generation system for recipe instructions, dish recommendation, and ingredient parsing.
- Processed over 300 structured Markdown documents and built a modular pipeline for data preparation, index construction, retrieval optimization, and LLM generation.
- Implemented a parent-child document retrieval strategy to reduce context fragmentation.
- Used fine-grained sub-chunks for vector search while passing complete parent documents to the LLM for better contextual integrity.
Awards and Honors
- National Scholarship, 2023-2024
- First Prize, 16th Chinese Mathematics Competitions (Non-Math Category A), 2024
- Second Prize, 14th Asia and Pacific Mathematical Contest in Modeling, 2024
- Second Prize, Pazhou Algorithm Competition (University Tour, SCUT Station), 2025
- Third Prize, 14th MathorCup Mathematical Modeling Challenge, 2024
- Third Prize, 13th Asia and Pacific Mathematical Contest in Modeling, 2023
Skills
- Languages: Chinese (Native), English (CET-4: 550+, CET-6: 620+)
- Research and engineering: LLMs, LoRA, Genetic Programming, RAG, PPO, surrogate models, path planning
- Programming: Python, C++, Markdown