Hi, there! 👋

A tinkerer, a bit geeky, and a programmer with too many ideas.

I'm Charles (aka 安妮的心动录), a full-stack engineer focused on Kubernetes, cloud-native infrastructure, and AI Agents. I can work end-to-end across frontend, backend, testing, deployment, and operations.

✍️ I share engineering practice, AI Agent notes, investing ideas, and personal insights on X, where I currently have 2K+ followers.
📈 I currently invest mainly in US stocks and Crypto, and I write about investment methods that ordinary people can actually execute.
💬 For consulting, add me on WeChat and make sure to include your purpose in the request; otherwise I will not accept it.

🎯 Hobbies

Travel, scuba diving, photography, hiking, reading, movies, music, investing. I prefer nature and also enjoy spending time alone.

🏠 Life

Living with my long-time partner and our cat Nuomi (糯米).

🌏 Visited

Experience

Tencent Music Entertainment 2023.05 — 2026.05

2023.05 — 2026.05

AI Platform Engineering

Led the development of an enterprise-wide AI deployment platform from scratch. Designed the complete architecture including dual-channel WeChat integrations (both personal and enterprise WeChat), cross-container persistent memory, a Skills marketplace with GitLab-based version control and rollback, and a Multi-Agent persona system. The platform reached widespread internal adoption within weeks of launch.

Release & Deployment Engine

Built a standardized release template system that powers configuration-driven deployments across all business lines. Designed a parent-child template architecture that supports API-driven publishing, drastically reducing manual deployment steps and enabling reproducible, auditable releases at scale.

Container Platform Management

Managed the group's large-scale container infrastructure spanning thousands of nodes and GPU cards. Designed and implemented intelligent node scheduling to balance cluster loads, built full pod lifecycle management with auto-scaling (HPA), timed expansion for traffic spikes, and automatic resource governance to prevent waste and reduce operational costs.

Baidu 2023.02 — 2023.05

2023.02 — 2023.05

Backend Development Intern — Baidu APP

Worked on the core interaction system for Baidu APP, including comments, likes, danmaku (bullet comments), and smart content moderation. Built a machine-driven comment insertion system with automated quality scoring. Refactored the intelligent review module using Go for better performance and maintainability. Developed a smart account operations system for content seeding, using Go backend with a low-code amis frontend.

TMLake (通明智云) 2022.09 — 2022.12

2022.09 — 2022.12

Container & Monitoring Development Intern

Focused on container orchestration tooling and observability infrastructure. Worked on containerized application deployment workflows and developed Prometheus-based monitoring solutions — including custom metric collection, alerting rules, and storage extension for long-term metric retention.

PANDAG (食铁兽科技) 2022.03 — 2022.09

2022.03 — 2022.09

Early Team Member

Joined as an early team member during sophomore year, working alongside senior schoolmates to build the company from the ground up. PANDAG has since secured investments from Jack Li (李泽湘), Miracle Plus (奇绩创坛) and other notable investors, successfully closing its Series A round and growing to nearly 100 employees.

Projects

talk-mbti

对话式 MBTI 人格分析 —— 没有一道选择题，打开就是聊天。心理学里的社会期许偏差（Social Desirability Bias）告诉我们，人填问卷时会无意识地美化自己。talk-mbti 换了个思路：通过自然对话，从你的措辞、第一反应、决策逻辑里提取真实的行为信号，聊完直接生成四维度百分比、认知功能栈、超能力和成长盲区的完整报告。基于 Gemini 2.0 Flash 驱动，Cloudflare AI Gateway 代理，国内无需梯子可用。

AIMBTIPrompt

hermes-agent-anatomy

Hermes Agent 源码解剖 —— NousResearch/hermes-agent 是一个 33 万行 Python 代码的开源 AI Agent 框架，从 CLI 交互到 11 平台消息网关到 RL 训练数据采集，全部内置在一个单体仓库里。这个项目是对其源码的系统化技术分析，逐模块拆解，写了 8 篇技术文档，配有大量手绘风格技术插图。从 Agent 核心循环、Tool Registry、多 Provider 适配到上下文压缩、消息网关、Memory 与 RL 训练。最后一篇将 Hermes Agent、OpenClaw、Claude Code 三个框架放在一起做了 13 个维度的对比。

AI Agent源码解析Hermes

claude-code-docs

Claude Code 源码解剖 —— 2026 年 3 月 Anthropic 意外泄露了 Claude Code 的完整 TypeScript 源码（51.5 万行）。我对这份源码做了系统化的技术分析，逐模块拆解，写了 13 篇技术文档（中英双语）。从架构设计到 Agent 循环的六阶段实现，从三层消息压缩体系到 6,300 行的权限系统，从五层记忆加载到 40+ 工具的执行流水线。同时从 82 个 feature flag 里挖出了 Anthropic 尚未发布的功能蓝图。

Claude Code源码解析AI Agent

claude-code-config

Claude Code 配置最佳实践 —— 装完 Claude Code，第一件事找别人的配置抄。settings.json、CLAUDE.md、Skills、MCP、Hooks 五个维度组合空间巨大，官方文档只告诉你每个选项是什么意思，不告诉你怎么配才好用。这份配置经过实战验证，一行命令安装，覆盖权限策略、安全兜底规则、效率优化等方面。从能用到好用，中间差的就是一份靠谱的配置。

Claude Code配置最佳实践