用 Jev 给 3518 份实习申请打分筛人
团队用 Jev 搭内部评审台，4 分 54 秒筛完 3518 份申请、约 0.41 美元，挑出 20 份给人看；作者结论是招聘变量太多，AI 只需在流程节点做对决策。
Jev is awesome but Recruiting is HARD! our team built an internal review console using Jev to help us make sense of a massive 4,000+ candidate pool at scale. Jev is interesting because it doesn’t try to generate another wall of text, it’s built to make fast, structured decisions that software can actually use. we gave it our criteria and let it score 3,518 applications: > 3,518 applications checked in 4:54 > 20 surfaced for human review (not the final list) > ~$0.41 estimated Jev cost every square is a real SDE intern application. and honestly, Jev was really good at the part we asked it to do. but recruiting is a different beast. there are way too many variables technical depth, ownership, trajectory, quality of work, context, potential, and the weird signals you only catch when you actually dig into someone’s work. the bigger takeaway for me: a lot of AI doesn’t need to generate anything. it just needs to make the right decision at the right point in a workflow. making those decisions fast and cheap changes what you can automate and that’s where i think Jev gets really interesting. https://x.com/idleshubh/status/2102429846660079971/video/1