关于 About
关于我
我是刘祚先，一名关注 AI Agent、复杂智能系统与软件工程实践的全栈开发者，拥有格拉斯哥大学计算机科学硕士学位。
我的技术经历从前后端与跨平台应用开发，逐步延伸到云原生系统、机器学习应用、AI Agent 和量化金融研究。我更关注如何把新技术转化为可运行、可验证并且能够持续演进的实际系统，而不只是完成孤立的功能或演示。
关注方向
目前，我主要关注以下方向：
- AI Agent 的工作流设计、执行审计、自我改进与工程化落地；
- AI 原生软件开发中的需求管理、架构控制、上下文管理与质量验证；
- 云原生环境中的服务编排、系统监控与自动化管理；
- 时间序列模型、量化金融和智能决策系统的实践探索。
这些方向并不是彼此独立的技术标签。我的兴趣在于理解它们如何共同构成复杂系统，以及如何在真实约束下保持系统的可靠性、可维护性与可解释性。
工程实践方法
我将 AI 视为工程协作者，而不是代替思考和责任的工具。
在实际项目中，我会让 AI 深度参与调研、设计、编码、测试和文档整理，同时由人负责明确需求、控制架构、验证结果并完成关键决策与 Review。相比单纯追求代码生成速度，我更加关注开发过程是否可追溯、结果是否能够被验证，以及系统是否仍然处于人的理解和掌控之中。
关于这个网站
这个网站用于记录我的工程实践、研究探索和技术思考。
这里既包括已经完成的项目和可复现的技术经验，也包括仍在进行的实验、失败的尝试以及对 AI 时代软件开发方式的反思。我希望这些内容不仅记录“做了什么”，也能说明“为什么这样做”“结果如何”以及“下一次会如何改进”。
联系方式
About Me
I am Zuoxian Liu, a full-stack developer focused on AI agents, complex intelligent systems, and practical software engineering. I hold an MSc in Computing Science from the University of Glasgow.
My experience began with frontend, backend, and cross-platform application development and has gradually expanded into cloud-native systems, applied machine learning, AI agents, and quantitative finance research. I am particularly interested in turning emerging technologies into systems that are operational, verifiable, and capable of evolving over time, rather than building isolated features or short-lived demonstrations.
Areas of Interest
My current areas of interest include:
- Workflow design, execution auditing, self-improvement, and production engineering for AI agents;
- Requirements management, architectural control, context management, and quality assurance in AI-native software development;
- Service orchestration, observability, and automated management in cloud-native environments;
- Practical exploration of time-series models, quantitative finance, and intelligent decision systems.
I do not see these as disconnected technology labels. I am interested in how they interact within complex systems and how such systems can remain reliable, maintainable, and explainable under real-world constraints.
Engineering Approach
I treat AI as an engineering collaborator, not as a substitute for human reasoning or responsibility.
In my projects, AI may participate extensively in research, design, implementation, testing, and documentation. Human oversight remains responsible for defining requirements, controlling architecture, validating results, and making critical decisions and reviews. Rather than pursuing code-generation speed alone, I care more about whether the development process is traceable, whether its results can be verified, and whether the system remains understandable and under human control.
About This Website
This website documents my engineering work, research explorations, and technical reflections.
It includes completed projects and reproducible technical experience, as well as ongoing experiments, unsuccessful attempts, and reflections on software development in the age of AI. My goal is not only to record what I built, but also to explain why particular decisions were made, what results they produced, and what I would improve next time.
Contact
- Curriculum Vitae
- GitHub
- International email: zuoxian.liu.work@gmail.com
- China email: liulzx2001@qq.com