Profile
Last updated: 2026/10/13.
I am a bachelor's graduate in Data Science and Big Data Technology from China University of Mining and Technology. My undergraduate thesis was an intelligent resume generation assistant based on large language models, supervised by Professor Xia Zhanguo. I focus my work and learning on AIGC + Fab, data analysis and visualization, semiconductor yield improvement, semiconductor AI Agents, and model fine-tuning. With a solid foundation in computer systems, I am committed to translating frontier AI technology into practical business value. I am currently employed at China Resources Microelectronics · Wuxi.
Education
B.Eng in Data Science & Big Data Technology (GPA: 3.44/4.0)
Work Experience
- Participated in Agent platform development, covering the yield knowledge hub, IT system construction for manufacturing-center modules, and FDC sign-off authorization for module equipment.
- Developed an industrial AI decision model for manufacturing-center PM business, applying reinforcement learning to post-train large models.
- Built web applications on the OPAS big-data platform, and developed data collection, intelligent monitoring, and email alerting systems for FDC, SPC, and KeyTool data.
- Developed Agent applications on Dify for business scenario deployment.
Skills
实习经历
- 参与大模型训练数据构建，针对计算机领域文本数据进行语义纠错与质量审核。
- 负责多模态业务场景实践，进行图像内容分类及音频 ASR 识别准确性校验。
项目经历
- 设计 8 个专业 Agent 协作，基于 LangGraph 状态图编排任务流转，并保留人工终审环节，使 AI 辅助决策始终可控、可追溯。
- 打通「AI 辅助决策 · 人工终审把关 · 全链路合规对接 · 数据安全可控」的完整闭环。
- 采用本地数据池 + 合规 API 对接：业务数据与向量数据本地留存，外部大模型能力经合规通道调用。
- 技术栈：前端 Vue 3 + Ant Design Vue 4 + LangGraph；后端 Python FastAPI + SQLAlchemy + Alembic，提供 97+ RESTful 接口；数据层 PostgreSQL 16 + Redis/Celery + Qdrant 向量库；AI 层 DeepSeek API + LangGraph 状态图 + CLIP 本地向量。
提示：以下为本科期间项目内容，未来会更新工程项目。
- 设计“端云混合”调用架构，基于 Qwen3.5-9B 模型结合 RAG 技术保障生成内容的专业性。
- 采用 BERT-Whitening 算法计算 JD 语义相似度，实现 Top-5 岗位匹配率 ≥ 82%。
- 利用灰狼优化算法 (GWO) 自动化优化网络超参数，将模型测试集准确率提升至 99.17%。
- 引入 Grad-CAM 实现决策可视化，通过生成热力图直观展示模型关注的故障振动区域。
- 基于 1980-2017 赛季 2835 名球员数据，通过 EDA 挖掘三分球比例提升等战术趋势。
- 构建 DNN 模型进行位置预测，准确率达 90.83%；利用 K-Means 聚类探索现代篮球新角色。
荣誉证书
- 全国三等奖 | 全国大学生计算机设计大赛 (2025.08)
- 全国三等奖 | 全国仿真创新应用大赛 (2024.12)
- 平煤神马企业奖学金 (2024)
- 校级二等奖学金 | 中国矿业大学 (2023, 2024, 2025)
自我评价
A data science graduate with a solid computer systems foundation, focused on applying AI, data analysis, and visualization to semiconductor manufacturing scenarios. I value rigorous problem decomposition, cross-functional collaboration, and turning model capabilities into measurable business outcomes.