Skip to content

wdkang123/wdkang123

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

16 Commits
 
 
 
 

Repository files navigation

武子康 | Zikang Wu

Backend Platform Engineering × AI Agent Systems × Edge-Cloud Multimodal Interaction

写代码,做系统,建工具,记录长期成长。

我有 5 年左右 Java 后端与平台研发经验,长期参与微服务平台、数据链路、高并发系统、性能优化、工程化交付和复杂业务系统建设。当前技术主线是以后端平台能力为底座,把 LLM、Agent、Function Call、模型服务、实时语音链路、业务 API 和边端设备整合成可部署、可评测、可运维的 AI 应用系统。

这里是我的 GitHub 入口;更完整的项目地图、技术文章、系列内容和长期记录,放在我的个人技术门户:

Personal Technical Portal: wzk.icu

Profile Snapshot

  • Backend platform foundation: Java、Spring Boot / Spring Cloud Alibaba、Nacos、Gateway、MyBatis、Spring Security、Kafka、Redis、Elasticsearch、MySQL、XXL-JOB、Docker、Kubernetes、Jenkins、Harbor、Rancher、EFK。
  • AI application engineering: LLM、ReAct、Function Call、MCP、Agent orchestration、model serving、local deployment、RAG、ASR、TTS、real-time voice pipelines、evaluation、observability。
  • Complex system integration: 擅长把模型能力、业务 API、数据链路、实时通信、权限治理、日志评测、部署交付和终端接入组织成完整闭环。
  • Career direction: AI Agent 工程师、多模态交互平台工程师、AI 应用基础设施工程师、后端平台研发工程师、端云协同与具身智能应用系统工程师。

Explore

  • Start Here — 按 AI Engineering、AI Infra、Backend Engineering、Indie Building、AI Writing Quality 和 Digital Life 选择阅读路径。
  • Project Map — 7 个项目档案,把个人技术门户、AI 工程化、独立工具、工程效率和数字生活实验串起来。
  • AI Engineering — 从 LLM、Agent / MCP、RAG、实时语音、模型服务和可观测性进入 AI 落地问题。
  • AscendLab Tools — 175 个免费 browser-side tools,覆盖开发、图片、视频、内容、金融、健康、数学和时间规划。

What I Am Building

  • wzk.icu — Personal Technical Portal / 个人技术门户 长期维护的技术档案、项目地图和写作入口,围绕后端工程、AI Engineering、独立构建、技术文章和长期成长记录展开。当前沉淀 655 篇技术文章、7 个项目档案、6 条长期主题路径,并通过 Sitemap、RSS、llms.txt 和结构化数据提升搜索与 AI 可读性。可以从 Start HereProjectsTopicsSeriesBlog 开始。

  • AscendLab — Browser Tools & AI Workflow Previews 一个独立产品实验,当前公开 175 个 browser-side free tools,覆盖 calculators、converters、developer utilities、image/video tools、content tools、finance / health / math helpers 和 time tools。Public tools are free and require no account; AI workflows are still preview / early access / planned, not a mature SaaS yet.

  • AI Voice Intelligence Module — 端云协同 AI voice interaction system 正在推进的 AI 工程化项目:端侧关注 wakeup、recording、VAD、playback、interruption、task trigger 和 status feedback;云侧串联 STT → LLM → tool calling → TTS,并通过 WebSocket / gRPC / MQ / Gateway、Agent / MCP、runtime config、trace / observability 把语音交互从 demo 推向可配置、可观测、可扩展的系统。长期方向是 one brain, multiple forms.

  • AI Infra Handbook — AI infrastructure learning and practice handbook 一个学习与实践手册,用来沉淀 LLM application architecture、RAG、Agent / MCP、model serving、local deployment、inference deployment、GPU、Kubernetes、evaluation、observability、latency optimization 和 cost governance。不包装成权威课程,更像持续更新的工程笔记。

  • stop-slop-zh — Chinese AI writing quality project 面向中文 AI 写作质量治理的规则项目。它不是反 AI,而是反对 vague、mechanical、template-like、fake-depth writing,重点服务中文博客优化、prompt engineering、内容审查和未来 Skill-style checks。

  • Code Generator — internal engineering productivity tool 内部工程提效实践:围绕 database / table selection、DTO / Model / VO / Service / DAO / Controller generation、template-based generation 和 AI-assisted API document parameter extraction,减少重复业务代码。生成代码仍然需要工程审查。

  • TeslaMate — digital life experiment 把 driving data、charging、energy cost 和 Grafana dashboard 组织成可回看的个人数据工程实验,用技术辅助日常生活复盘。

  • Robotics & Embodied AI Application Layer — 机器人 AI 交互与端云协同 关注机器人、巡逻车、Jetson Nano、ESP32 等终端接入,把自然语言指令、业务 API、设备动作、状态回传和语音反馈组织成应用系统。重点不在底层 SLAM、运动控制或强化学习算法,而在 AI 交互、任务编排、终端接入和动作安全门控。

Technical Stack

Backend & Platform Engineering

Java、Spring Boot / Spring Cloud Alibaba、Nacos、Gateway、MyBatis / MyBatis-Plus、Spring Security、MySQL、Redis、Elasticsearch、RabbitMQ / RocketMQ / Kafka、XXL-JOB、REST APIs、microservice governance、authentication、tracing、Docker、Kubernetes、Jenkins、Harbor、Rancher、EFK、CI/CD、DevOps。

AI Engineering

LLM application engineering、ReAct、Function Call、Agent / MCP / tool calling、RAG、model serving、vLLM、local deployment、STT / LLM / TTS pipelines、real-time voice interaction、WebSocket / gRPC、runtime configuration、evaluation、trace and observability。

Frontend & Product Building

TypeScript、React / Vue、Next.js / Astro、static site generation、SEO / GEO、browser-side tools、product documentation。

Automation & Writing

Codex / Claude Code / ChatGPT-style AI workflows、prompt engineering、technical blog optimization、Markdown documentation、project review、long-term writing systems。

What I Write About

  • 后端工程、微服务治理、平台研发、数据链路和稳定性交付。
  • LLM 应用工程化、Agent / MCP、Function Call、RAG、模型服务、本地化部署、实时语音链路、AI Infra 和可观测性。
  • 端云协同、多模态交互、机器人 AI 应用系统和动作安全门控。
  • 独立工具、浏览器工具、个人自动化、SEO / GEO 和公开作品系统。
  • 中文 AI 写作质量治理、博客优化、项目复盘和长期成长记录。

Current Focus

  • wzk.icu 维护成可搜索、可回看、可被 AI 理解的个人技术门户。
  • 持续整理 AscendLab 的免费浏览器工具和 AI Workflow preview。
  • 推进 AI 语音智能模块的 trace、latency、runtime config、tool calling、router、streaming output 和端到端稳定性。
  • 沉淀 AI Infra Handbook,把 AI 应用背后的基础设施知识整理成可复用地图。
  • 探索端云协同多模态 Agent:把 Web 平台、机器人、巡逻车、模型服务和业务系统接入统一交互中枢。
  • 维护 stop-slop-zh,把中文 AI 写作质量问题转成可检查、可协作的规则。

Principles

  • 先做成能运行的系统,再谈抽象。
  • 不把 demo 当产品,不把概念当结果。
  • 工程输出需要可复盘、可观测、可迭代。
  • 不让 LLM 直接决定高风险设备动作,动作下发前必须经过环境识别、规则约束和安全门控。
  • 技术栈只是手段,真正重要的是对问题的理解和解决。
  • 长期记录比短期声量更重要。

Public Links


我关注的是一条连续的工程主线:用后端平台搭底座,用 AI Agent 和模型服务扩展能力,用端云协同与独立构建把想法变成可落地、可复盘、可长期演进的系统。

About

武子康

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors