How I Learned to Customize My Resume for Every Job (and Built an AI Tool to Do It)

#ai #career #productivity #webdev

When I was younger, I applied for jobs the wrong way. My CV looked creative — boxed sections, custom layouts, a thoughtfully crafted one-size-fits-all version — and my cover letters were carefully written with original wording instead of copying phrases from the job description. I thought this showed effort and seriousness.

What I didn’t realize was that Applicant Tracking Systems (ATS) couldn’t parse my resume properly. Most employers never even saw it. My cover letters were never read either. At the time, I had no idea why my job search kept failing.

One day, I learned something embarrassingly simple: my resume format was wrong, and it lacked keywords from the job description. Once I fixed that and started aligning my resume with the roles I was applying for, my response rate changed dramatically. I started getting replies — something that almost never happened before. That moment planted a seed.

Instead of stopping there, I convinced myself there was something bigger behind this insight. That’s how the idea for ResumeCustomizer was born: a tool that generates a custom resume for every job application.

I started building it before I had real web development experience — and long before AI was usable for tasks like this. Angular and React were just emerging at the time. I chose Angular and struggled through the learning curve.

Eventually, I built a basic MEAN-stack application that accepted personal details and a job description, stored data in MongoDB, and generated DOCX and PDF resumes.

The customization logic was very simple. It looped through every word in the job description and filtered out the 1,000 most common English words. What remained were usually key technologies and role-specific keywords. The user cleaned that list manually, and those keywords were injected into the skills and personal profile sections. It wasn’t great — but it worked for simple IT roles.

I also experimented with TF-IDF, but it required a large document corpus and became too complex to properly implement. None of these early versions ever went live. Still, the idea never left me.

The modern version

Today ResumeCustomizer is rebuilt with:

Next.js 15 (App Router)
React
Tailwind CSS + shadcn/ui
Prisma + PostgreSQL
Clerk for authentication
Stripe for payments
ChatGPT API for resume customization
Deployed on Google Cloud Run
Managed with Bun

AI now handles:

Keyword extraction from job descriptions
Personal profile generation using those keywords
Enhancement of job bullets using some of these keywords

The process of generating a custom resume behind ResumeCustomizer is intuitive and straightforward. It is also discussed in more detail in my Reddit post here