I built an open-source bot that applied to 8,202 Dice.com jobs in 10 days (Python + Playwright)

reddit SideProject · 原文链接 ↗ · 60 分

I'm a senior AI/ML engineer who works contract. When you're on the bench, applying becomes a full-time job: the same role gets posted by 15 different vendors, and visibility only comes from volume. So I automated the volume part and kept the human part for the calls. What it actually does Logs into Dice, cycles through keywords you control ( keywords.txt ), clicks Easy Apply on every match Logs every application to applications.csv and skips anything already in there, so it never double-applies Sleeps 2-3 AM and wakes back up on its own. My longest run was 10 days unattended

五维评分

AI 分析

痛点：合同制技术岗位求职者需要海量投递才能获得可见度，但重复职位多、表单繁琐，人工投递耗时且易漏投/重复投。

目标用户：合同制/自由职业的技术求职者、需要批量投递的猎头和招聘代理。

证据：When you're on the bench, applying becomes a full-time job: the same role gets posted by 15 different vendors, and visibility only comes from volume. ... 8,202 applications submitted, 99.76% success, 10 failures, 0 CAPTCHAs or bot checks, ~250-300 per day, 10 days unattended

变现：开源核心 + 托管 SaaS 订阅：按投递量/账号数/职位板收费，高级功能包括简历定制、表单识别、分析报表、多账号隔离、Telegram/邮件提醒；面向猎头/代理提供团队席位。

MVP：2周内做一个单职位板（如 Dice 或 Greenhouse/Lever）的浏览器扩展/桌面工具：关键词配置、自动 Easy Apply、去重 CSV、答案映射、人工复核队列、每日摘要。不绕过 CAPTCHA，不猜答案，先服务自己或小范围用户。

竞品：LazyApply、Sonara、Simplify、LoopCV、AIHawk、Jobscan

摘要：一位合同制 AI/ML 工程师开源了一个 Python + Playwright 机器人，在 Dice.com 上 10 天自动投递 8,202 个职位，成功率 99.76%，每天约 250-300 个，无 CAPTCHA。痛点真实：bench 期投递像全职工作，同一职位被 15 个 vendor 重复发布，只能靠走量获得可见度。但项目已开源且竞品众多，变现需靠托管服务、高级功能或面向猎头/代理的团队版，付费意愿未在原帖中直接验证。