Front-End Isn't Just Slicing Mockups: In the AI Era, People Who Understand Fundamentals Can Go Full Stack End to End
Supports switching languages:
简体中文- 繁体中文
English
A front-end developer's honest takeaway after building an entire enterprise system alone: AI can boost efficiency dramatically, but the real advantage still comes from understanding fundamentals and working across the stack.
Recently, I finished an enterprise-grade system piece by piece. I built the front end, back end, database, cache, and file storage myself. AI definitely helped a lot during the process, but my takeaway after finishing is a bit different from what many people imagine.
Let's start with the stack, no mystery here:
Front End
- 1. Build pages with Nuxt 3 + Vue 3
- 2. Use Element Plus for components and Pinia for state management
- 3. Use ECharts for reports and charts
Back End
- 1. Build services with NestJS. Its layered structure actually feels a lot like writing Vue components, so the jump from front-end is not that hard.
- 2. Use Sequelize + MySQL for database operations
- 3. Use Redis for caching. Permissions and hot data go through cache, which makes APIs much faster
- 4. Store files and attachments with S3 / Alibaba Cloud OSS
- 5. Use JWT for login authentication, and Swagger to auto-generate API docs
What felt best to me is this: from front end to back end, the core is all JavaScript, so I don't need to constantly switch my mental model.
Back to AI. It really saved me a lot of time. For repetitive code, docs lookup, and boilerplate, it's faster than I am. But there were several times when its solution looked right in logic, yet failed at runtime, or buried performance pitfalls. If I didn't understand the underlying principles, I would never have caught those issues and would have just accepted everything at face value.
So my view now is: AI doesn't mean you can stop learning. It means people who understand the fundamentals can move faster. The more you know, the better you can judge whether what it gives you is correct, good, and usable. If you don't understand it, AI can easily lead you in the wrong direction.
If you're still wondering whether it's worth learning now that AI is here, my answer is yes. You don't need to panic about mastering everything, but the underlying logic is worth your time. That's the part AI can't take away.
If you're also self-learning full stack like I am, drop a comment and let's connect over source code.
最近断断续续做完了一套企业级系统，前端、后端、数据库、缓存、文件存储都是自己搭的。整个过程确实用了不少 AI 辅助，但做完之后的感受，和很多人想的不太一样。
先说技术栈，没有藏着掖着：
前端
- 1. Nuxt 3 + Vue 3 写页面
- 2. Element Plus 做组件，Pinia 管状态
- 3. ECharts 出报表图表
后端
- 1. NestJS 写服务，分层其实和写 Vue 组件挺像，前端转过来不算难
- 2. Sequelize + MySQL 操作数据库
- 3. Redis 做缓存，权限和热点数据走缓存，接口快很多
- 4. 文件、附件用 S3 / 阿里云 OSS 存
- 5. JWT 做登录鉴权，Swagger 自动生成接口文档
最让我舒服的一点是：从前端到后端，核心都是 JavaScript，不用频繁切换思维。
说回 AI。它确实帮我省了很多时间，重复代码、查文档、写样板，它比我快。但有几次它给的方案是有问题的，逻辑看着对，跑起来不对，或者性能埋了坑。如果我不懂背后的原理，这些坑我根本发现不了，只会照单全收。
所以我现在的看法是：AI 不是让你不用学，而是让懂原理的人跑得更快。你懂得越多，越能判断它给的东西对不对、好不好、能不能用。不懂的话，它反而容易把你带偏。
如果你也在犹豫「有 AI 了还要不要学」，我的答案是：学。不用焦虑要全都会，但底层逻辑值得花时间。这部分东西，才是 AI 拿不走的。
和我一样在自学全栈的，评论区交个朋友交流源码。
最近斷斷續續做完了一套企業級系統，前端、後端、數據庫、緩存、文件存儲都是自己搭的。整個過程確實用了不少 AI 輔助，但做完之後的感受，和很多人想的不太一樣。
先說技術棧，沒有藏著掖著：
前端
- 1. Nuxt 3 + Vue 3 寫頁面
- 2. Element Plus 做組件，Pinia 管狀態
- 3. ECharts 出報表圖表
後端
- 1. NestJS 寫服務，分層其實和寫 Vue 組件挺像，前端轉過來不算難
- 2. Sequelize + MySQL 操作數據庫
- 3. Redis 做緩存，權限和熱點數據走緩存，接口快很多
- 4. 文件、附件用 S3 / 阿里雲 OSS 存
- 5. JWT 做登錄鑑權，Swagger 自動生成接口文檔
最讓我舒服的一點是：從前端到後端，核心都是 JavaScript，不用頻繁切換思維。
說回 AI。它確實幫我省了很多時間，重複代碼、查文檔、寫樣板，它比我快。但有幾次它給的方案是有問題的，邏輯看著對，跑起來不對，或者性能埋了坑。如果我不懂背後的原理，這些坑我根本發現不了，只會照單全收。
所以我現在的看法是：AI 不是讓你不用學，而是讓懂原理的人跑得更快。你懂得越多，越能判斷它給的東西對不對、好不好、能不能用。不懂的話，它反而容易把你帶偏。
如果你也在猶豫「有 AI 了還要不要學」，我的答案是：學。不用焦慮要全都會，但底層邏輯值得花時間。這部分東西，才是 AI 拿不走的。
和我一樣在自學全棧的，評論區交個朋友交流源碼。
