Jin Yoo

I got laid off. So I built a Job Tracker & Resume Builder powered by LLMs, using LLMs.

On Thursday Feb 26, Block laid off nearly half of its employees and I was impacted by it.

As I was cleaning the dust off my job seeker hat, I found myself annoyed at the prospect of setting up a spreadsheet to track job applications and managing multiple versions of my resume. But wait, I'm an Engineer. We invest hours building and automating things, when doing the actual thing would have taken us much less time. And I'm also an "AI Engineer" now, remember? I have become so much more productive at building software using AI, and I have a deep interest in building software that uses AI.

That is why I built a Job Tracker & Resume Builder that uses LLMs, using LLMs.

You can find the code and instructions at https://github.com/fiffty/JobCinch, with videos illustrating the features.

What the app is

JobCinch is an AI-powered tool for job seekers that combines job tracking with resume building. Drop a link to a job posting and JobCinch will parse and save it, letting you manage your applications and track their status in one place. For resumes, import your existing resume to build a master profile that accumulates all your experience, then use AI to generate tailored resumes optimized for specific job postings.

What makes this app unusual is that it's designed to be used alongside AI coding tools like Cursor or Claude Code. This was my thought process that got me to this idea:

  1. I want an app where I can track job postings, but with niche features. For example, I'm searching jobs across multiple countries, so I'd like to compare compensation values in the same currency. But I couldn't find something that had this.
  2. I also need to brush up on my resume. I need to update my master resume, and start to tailor it to individual job postings. I used to do this by hand, but now I can probably lean on Gemini in Google Drive or something.
  3. Wait, what if I build a job tracking app just for myself, with features unique to my use case. Building apps of small scope has become super easy and fast.
  4. For resumes, I could copy my master resume over from Google Doc into something like a markdown document. Then I can set up some commands or subagents in AI coding tools to set up reusable flows to create tailored resumes.
  5. Or what if I pulled the resume related features into the job tracking app that I'm trying to build. But that would involve having to build an agent framework around making API calls to LLM providers, and that seems like too much work.
  6. The lightbulb moment. What if I build an app with the explicit expectation of using it alongside other AI coding tools? Instead of persisting jobs and resume data in a database, I can have them as files for the AI tools to read and update.

The most powerful feature is the /create_job_resume flow. It takes your master resume and reviews it against a job posting. First it creates a new resume with only the relevant bullet points. Then it identifies gaps -- what the company wants from a candidate but isn't showcased in your resume -- and engages you in a back and forth conversation to build more bullet points. Once all of the gaps have been discussed, it creates the final version of your tailored resume and updates your master resume with the new content so that future runs can use them. Here is a quick demo of it:

Local, personalized, purpose-built

It has become genuinely easier and faster to build an application, especially one of smaller scope. A few weeks ago, Anthropic's Claude Cowork caused a software stocks selloff, and while stocks saw some rebound as Anthropic partnered with SaaS companies, the writing is on the wall. The barrier to building custom software is dropping fast.

I would even argue that small homegrown apps have an edge over large SaaS offerings in the area of "features that matter". Haven't we all had experience stressing out on picking an app or SaaS service to use, because none have the exact combination of features that I am looking for? Not all apps need scale, not all apps need the network effect. Many apps, instead, would benefit greatly with smaller scale, smaller scope, in exchange for a deeply personalized set of features.

I don't know if building small personal apps will remain a hobbyist trend or become something much bigger. But what I do know is that I will continue to build. Constantly.

Some other parting thoughts