纯前端工具Frontend-only tool· 2026
JobMatch
隐私优先的岗位筛选工具Privacy-first job filtering tool
零后端、零登录、零隐私上传，在浏览器内完成 Excel 岗位表筛选。A browser-only Excel job filter that keeps sensitive personal conditions local while helping users compare public-sector roles.
案例摘要Case summary
目标用户Audience
担心隐私泄露、需要快速筛选岗位的考公考编用户。Civil-service exam candidates worried about privacy leaks who need to quickly filter positions.
我的角色Role
负责产品定位、信息架构、隐私边界和导出流程的产品决策与验收；全部代码由 AI 编程工具完成。Owned product positioning, information architecture, privacy boundaries, and export-flow decisions; all code was produced with AI coding tools.
交付结果Outcome
把手工筛表流程压缩成浏览器内上传、合并、筛选、导出的闭环。Compressed the manual spreadsheet-filtering workflow into a closed loop of upload, merge, filter, and export in the browser.
产品洞察Product insight
考生最担心的不是不会筛岗位，而是个人信息被培训机构或陌生平台获取。把所有计算放在浏览器内，隐私本身就是产品价值。What candidates fear most is not failing to filter positions, but their personal information being harvested by training institutions or unfamiliar platforms. Keeping all computation in the browser makes privacy itself the product value.
项目简介Overview
JobMatch 面向考公考编用户，支持上传岗位表和报名统计表，在浏览器内完成解析、合并、筛选和导出。核心卖点不是算法炫技，而是敏感个人条件不离开本地。JobMatch targets civil-service exam candidates, letting them upload a position table and a registration statistics table, then parse, merge, filter, and export entirely in the browser. The core selling point is not algorithmic flashiness — it is that sensitive personal conditions never leave the local machine.
技术栈Tech stack
结果证据Evidence
在线 Demo 已部署到 Vercel，用户无需登录即可进入核心流程。A live demo is deployed on Vercel — users can reach the core flow without logging in.
public/screenshots 保留筛选和结果页截图，能证明完整操作闭环。public/screenshots keeps filtering and result screenshots that prove the complete operating loop.
Excel 解析、筛选和导出逻辑在浏览器端完成，个人数据不上传服务器。Excel parsing, filtering, and export all run in the browser; personal data is never uploaded to a server.
项目截图Screenshots
关键决策Key decisions
官方 Excel 表头命名不统一，精确匹配很容易失效。Official Excel headers are named inconsistently, so exact matching fails easily.
用同义词映射、fuse.js 模糊匹配和内容特征识别做三层兜底。Use three fallback layers: synonym mapping, fuse.js fuzzy matching, and content-feature recognition.
左右分栏布局在笔记本上割裂操作流。The left-right split layout broke the operating flow on laptops.
重构为从上到下的一次性流程，让用户按自然顺序完成筛选。Refactor into a top-to-bottom one-pass flow so users filter in a natural order.
想看完整的代码和实现细节？Want the full code and implementation details?
在 GitHub 上查看View on GitHub