Builder · Open-source Explorer · AI, Software & Products
Explore ideas. Build useful things. Share what I learn.
持续探索，把有价值的想法做成可验证、可复用、可分享的成果。
I have a master's background in Electronics & Communication Engineering and experience in machine learning and applied AI. I use GitHub as a public workshop for projects, experiments, engineering tools, and lessons learned—not just as a résumé.
我关注的不只有某个模型或岗位，而是如何从一个真实问题出发，经过研究、实验、构建与验证，最终形成别人也能理解、运行和继续改进的成果。
Explore repositories · Recent activity
- Applied AI & intelligent applications — LLMs, retrieval, RAG, evaluation, multimodal interfaces, and domain-specific AI systems.
- Agents & automation — tool use, stateful workflows, human-in-the-loop, durable execution, and AI-assisted software development.
- Developer tools & infrastructure — MCP ecosystem experiments, evidence-based verification, reliability, observability, and deployment.
- Research-to-building & creative products — machine learning, recommendation, reproducible experiments, content workflows, and product prototypes.
这些是当前的探索方向，不是未来的边界。工具、框架和项目会变，持续学习、实践与开放分享是主线。
| Project | Focus |
|---|---|
| 🧴 Beauty Industry RAG QA System | 化妆品知识问答的独立开源参考实现：混合检索、两级重排、证据门控、权限隔离和知识版本管理。 |
| 🤖 Customer Service AI Agent | LangGraph 多智能体客服参考实现：工作流编排、持久化状态、异步任务、人工审批与工具幂等。 |
| Project | Focus |
|---|---|
| 🛡️ MCP Evidence Gate | 实验性 MCP 安全扫描证据校验与发布准入工具；不等同于漏洞扫描器或官方 MCP 项目。 |
| 🧩 Agent Control Plane | 架构探索中：AI Coding 多 Agent 的任务委派、隔离执行、证据收集与精确版本验收。 |
| Project | Focus |
|---|---|
| 🎬 X-SuperPlay ContentOps | 从真实项目、实验和研究素材出发，探索可追溯的视频内容生产与质量检查工作流。 |
| 🌐 X-SuperPlay Global Edge | 参考实现 / 实验项目：自托管边缘网络、只读健康检查与桌面观测工具。 |
A note on project status: Some repositories are reference implementations or ongoing experiments, not production services. Their READMEs, tests, and evidence documents define what has actually been built and verified. No employer-owned production code or private business data is represented here.
Curiosity / Real problem
↓
Research & small experiments
↓
Working prototype
↓
Tests, evaluation & evidence
↓
Documentation & open sharing
↓
Feedback → next iteration
I value clear problem framing, reproducible results, honest limitations, and useful artifacts. A working prototype is a step forward; a verified, maintainable project is a different milestone.
I enjoy discussing applied AI, developer tooling, open-source experiments, and ideas that cross research, engineering, and products.
欢迎通过相关仓库的 Issues / Pull Requests 交流问题、提出改进或参与协作。
Browse the work: github.com/Xander-Xai.
Stay curious. Keep building. Share what you learn.