架构师 / 技术负责人 · Go 后端 · 云原生 · AI Agent
在智能硬件公司带一支软件研发团队,日常做架构设计、技术选型、核心功能开发, 以及把需求拆成任务、盯进度和验收。周末把这些年攒下来的后端经验做成开源项目。
我不太相信"选型要靠直觉",更喜欢先跑通一遍、量出真实占用和吞吐,再决定用不用。
| 项目 | 一句话定位 | 语言 | Stars |
|---|---|---|---|
| RuoYi-Go | 用 DDD 六边形架构重写的 RuoYi 后端(Iris + Gorm),前端配 RuoYi-Vue3 | Go | |
| SaaS-Zero | 多租户 Go 微服务底座:go-zero + ent + Casbin,开箱带网关 / 认证 / 基础数据 | Go | |
| mini-ruoyi | 1 核 1G 跑得动的低占用后端,出海项目的服务器开销不再是大头 | Go |
三套 Go 后端怎么选(按你手上的机器和项目体量):
- 想学 DDD 分层 / 六边形架构怎么落到 Go 里 → RuoYi-Go
- 要做多租户的大项目,需要网关、认证、基础数据一次到位 → SaaS-Zero
- 只有 1 核 1G 的小机器(个人站、出海小型服务) → mini-ruoyi
三份 README 里放了同一张对照表,从任一个仓库进去都能对上。
语言 · Go · Java · Kotlin · TypeScript / JavaScript · Python · Swift
后端 · Iris · Gorm · go-zero · ent · Casbin · gRPC · RESTful API · PostgreSQL · Redis
前端 · Vue 3 · Element Plus · React · Ant Design Pro · Umi · Svelte
基础设施 · Linux · Docker · Kubernetes / K3s · Nginx · Jenkins · Git
移动端 · Android(Jetpack Compose) · iOS
AI / 端侧 · PyTorch · ONNX · RKNN · Rockchip NPU · FunASR · 语音唤醒 · 说话人分离
智能硬件 · IoT · 音视频(WebRTC / P2P / LiveKit) · 全球云平台 · 端侧 AI 推理 · AI Agent(LLM / RAG / MCP)
长期的技术积累基本都围着这几条线走,开源的几个后端底座也是从里面的真实需求里长出来的。
- 架构优先,但要用数据说话 —— 先量真实占用和性能,再定技术方案
- 控制技术债 —— 能简单就不复杂,宁少不滥
- 对齐竞品 —— 主动找功能缺口,进 Backlog,而不是等别人提
- 进度可见 —— 任务拆分、每日更新、明确验收标准
- GitHub:@Kun-GitHub
- Email:hot_kun@hotmail.com
English
Architect & Tech Lead · Go backend · Cloud Native · AI Agents
I lead a software team at a smart-hardware company — architecture design, tech selection, core development, and turning requirements into tasks I then track through to acceptance. On weekends I turn years of accumulated backend experience into open-source projects.
I don't really buy "pick the stack by gut feeling" — I'd rather build it once, measure the real footprint and throughput, and then decide whether to use it.
Open Source
| Project | What it is | Lang | Stars |
|---|---|---|---|
| RuoYi-Go | The RuoYi backend rewritten with DDD / hexagonal architecture (Iris + Gorm), paired with RuoYi-Vue3 | Go | |
| SaaS-Zero | Multi-tenant Go microservice base: go-zero + ent + Casbin — gateway / auth / basedata out of the box | Go | |
| mini-ruoyi | A low-footprint backend that actually runs on a 1-core / 1GB box — server cost stops being the big line item | Go |
Which of the three Go backends to pick (depends on your machine and project size):
-
Want to see how DDD layering / hexagonal architecture actually lands in Go → RuoYi-Go
-
Building a multi-tenant project that needs gateway, auth and basedata in one go → SaaS-Zero
-
Only have a 1-core / 1GB box (personal sites, small overseas services) → mini-ruoyi
All three READMEs carry the same comparison table, so whichever entry point you land on lines up.
Tech
Languages · Go · Java · Kotlin · TypeScript / JavaScript · Python · Swift
Backend · Iris · Gorm · go-zero · ent · Casbin · gRPC · RESTful API · PostgreSQL · Redis
Frontend · Vue 3 · Element Plus · React · Ant Design Pro · Umi · Svelte
Infrastructure · Linux · Docker · Kubernetes / K3s · Nginx · Jenkins · Git
Mobile · Android (Jetpack Compose) · iOS
AI & on-device · PyTorch · ONNX · RKNN · Rockchip NPU · FunASR · wake word · speaker diarization
Where I'm heading — Smart hardware · IoT · Real-time communication (WebRTC / P2P / LiveKit) · Global cloud platforms · On-device AI inference · AI agents (LLM / RAG / MCP)
My long-term work sits on these lines, and the open-source backends above grew out of real needs inside them.
How I work
-
Architecture first, but let the data talk — measure real footprint and performance before locking a design
-
Keep tech debt low — simple over clever; less, not more
-
Benchmark competitors — go find the feature gaps and put them in a Backlog instead of waiting to be asked
-
Keep progress visible — split the work, update daily, define acceptance criteria
📫 @Kun-GitHub · hot_kun@hotmail.com
Get the real thing running first. Architecture can come after.
先把真实的东西跑起来,再谈架构。