-
🏫 教育背景：
- 哈尔滨工业大学 2025～now 软件工程硕士在读，校内研究方向：大模型在 SysML 中的应用（LLM + SysML）
- 中国农业大学 2024～2025 数据科学与大数据技术 第二学士
- 东华大学 2020～2024 旅游管理 学士
- 期间实习：Daxue Consulting（咨询项目的数据洞察）、Volvo（沃尔沃，供应链数据分析）
-
💡 职业方向：大模型的后训练（微调/蒸馏/量化/模型优化）与 Agent 算法开发（Agent 框架与交互策略）
-
🛠️ 当前目标：把研究和工程结合，做可落地的 LLM 后训练、Agent 系统与相关算法实现
-
🏸 业余爱好：羽毛球 / 摄影 / 音乐 / 背包旅行
专注于支持“大模型后训练 + Agent 算法开发”的技术栈，包含但不限于：
- 模型开发与训练：Python，PyTorch，Hugging Face Transformers，PEFT / LoRA，DeepSpeed，Accelerate
- 模型优化与推理：量化（4-bit/8-bit），bitsandbytes，ONNX，xFormers，模型蒸馏
- 分布式与加速：CUDA，NCCL，多卡训练，ZeRO，Ray / TorchElastic
- Agent 与强化学习：LangChain，RLHF / TRL，PPO / SFT，Agent 交互设计与策略工程
- 数据与实验管理：Pandas，Apache Spark，Datasets（Hugging Face），DVC，MLflow，Weights & Biases
- 部署与工程化：FastAPI，Gradio，Streamlit，Docker，Kubernetes，CI/CD
- 工具链与工程习惯：Git，Linux，Shell，性能调优与监控
- 网站: UrgencyWu.github.io
- Email（已隐写）: shaw-wu [at] qq.com （网站主页提供“显示邮箱 / 复制邮箱”交互按钮）
- GitHub: @UrgencyWu
- Masters (Software Engineering) student at Harbin Institute of Technology (2025–now). Research focus: application of large language models (LLM) in SysML.
- Second Bachelor's in Data Science & Big Data Technology, China Agricultural University (2024–2025).
- Bachelor's in Tourism Management, Donghua University (2020–2024).
- Internships: Daxue Consulting (data insights for consulting projects), Volvo (supply chain data analysis).
Post-training for large models (fine-tuning / distillation / quantization / optimization) and Agent algorithm & system development.
Focused on the stack needed for "post-training + Agent development":
- Model development & training: Python, PyTorch, Hugging Face Transformers, PEFT / LoRA, DeepSpeed, Accelerate
- Model optimization & inference: quantization (4-bit/8-bit), bitsandbytes, ONNX, xFormers, model distillation
- Distributed & acceleration: CUDA, NCCL, multi-GPU training, ZeRO, Ray / TorchElastic
- Agents & RL: LangChain, RLHF / TRL, PPO / SFT, agent interaction & strategy engineering
- Data & experiment management: Pandas, Apache Spark, Datasets (Hugging Face), DVC, MLflow, Weights & Biases
- Deployment & engineering: FastAPI, Gradio, Streamlit, Docker, Kubernetes, CI/CD
- Tooling & engineering practices: Git, Linux, Shell, performance tuning & monitoring
If you want a concise English-only CV blurb or a version with skill-level tags (Proficient / Familiar / Learning), tell me which format you prefer and I will add it.