Shao Meng: The Claude Code era still needs senior front-end engineers
English text is being prepared.
Citing Lovart as an example, Shao Meng pointed out that in PGC+UGC communities, covers uploaded by users range from small images of a few KB to ultra-high-definition large images of tens of MB, and non-developers often do not proactively pay attention to performance. He believes that the Skills of Claude Code / Codex focus more on code architecture performance and do not cover image and video resource issues, so senior engineers who can identify and drive performance issues across teams will only become more important in the AI programming era.
Topics
View original postYou say front-end is dead? As long as you have Codex / Claude Code, anyone with hands can build a website? Then could I ask you "AI Native" product managers and designers to remember to pay attention to performance when doing Vibe Frontend Coding 😂 Performance score 47, piles of large gifs and large pngs, and many more issues that don't even need a deep look—simply a negative example of front-end performance ... https://t.co/ElOmUw8pu5
View original
听你们说我们前端已死？ 只要有 Codex / Claude Code，有手就能做网站？那能不能拜托你们这些 “AI Native” 的产品经理和设计师们，Vibe Frontend Coding 时，记得关注一下性能 😂 Performance 47 分、成堆的大 gif、大 png，更多问题都不用深看，简直前端性能反面典型 ... https://t.co/ElOmUw8pu5
Body
首先：对不起 Lovart，我只是拿你来举个例子 😄
之所以用 Lovart 主要是因为它的产品性质：PGC + UGC 的类社区和模板形态，用户使用的封面可能千差万别，有可能是几 KB 的模糊小图，也可能是几十 MB 的超清大图，几万 * 几万那种... 设计和运营团队也不例外的会出现后者的问题，从设计软件原图导出直接怼进运营后台...
这时如果你是一个完全不懂前端工程的非研发人员，我真的不太相信你会有意识的关注性能问题，特别在前期各种数据都在 mock 阶段还很理想时。
再进一步，即使你的 Claude Code / Codex 本身有关注前端性能方面的 Skills，它也会更关注代码架构本身的性能，对于未来可能会产生的各种图像视频资源，覆盖不到。
再进一步，即使是专业前端，他对这类资源问题也有经验和意识，也可能因为与产品设计运营团队的沟通不畅，或者性能方面问题要为效果让步，而让这个问题被略过，直到大面积产生客诉，或者老板自己用一下发现卡，才被发现 😄
所以，在团队里，有一个在工程方面经验丰富，能够发现问题，愿意在跨团队时提出和推进问题解决的工程师，在 Claude Code / Codex 的时代，仍然很重要，而且只会越来越重要！