I Spent Two Days Screening CVs. Then I Decided to Build an AI to Do It in Two Minutes

I Spent Two Days Screening CVs. Then I Decided to Build an AI to Do It in Two Minutes

AI
Laravel
React

I think every developer has a project that starts not with a grand vision, but with a deep, personal frustration. For me, that project is Sivy. And it all began with a pile of PDFs.

A few months ago, a friend running a growing startup shared a problem. They had a single opening for a "Backend Developer" and received over 200 applications in three days. Their small HR team was completely overwhelmed. Out of curiosity, I offered to help screen them.

What I thought would be an interesting task quickly became a soul-crushing exercise in manual labor.

The Manual, Soul-Crushing Process of Screening CVs

For two full days, my world shrank to a dual-screen setup: a PDF reader on one side, and a spreadsheet on the other. The process was painfully linear and riddled with inefficiencies:

  1. The Keyword Hunt: I found myself scanning for the same keywords over and over: "Laravel," "PostgreSQL," "API," "AWS." It was a human version of Ctrl+F, but slower and far more prone to error.

  2. The Subjectivity Trap: After the 50th CV, fatigue set in. Did I give candidate #57 a fair chance, or was I just tired? Did my personal bias for clean CV layouts influence my judgment? The answer was almost certainly yes.

  3. The Graveyard of Lost Talent: I realized that many CVs were being discarded not because the candidates were unqualified, but because their experience didn't perfectly match the current role. A fantastic DevOps engineer's CV was in the pile, but because we weren't hiring for that role right now, their profile was effectively lost—buried in an email inbox, never to be seen again.

The "What If...?" Moment

On the evening of the second day, staring at the spreadsheet, a question hit me: What if we could treat every CV not as a disposable document, but as a structured piece of data?

What if, instead of a human manually scanning for keywords, a machine could:

  • Read and understand every CV instantly?
  • Score and rank every candidate objectively against a specific set of requirements?
  • Build a living, searchable database of all talent that ever applied, ready for any future role?

The goal wasn't to replace the HR team. The goal was to give them a superpower. To free them from the monotonous, low-value work of manual screening so they could focus on the high-value work of actually talking to the best candidates.

Introducing Sivy: My Answer to the Chaos

That's the motivation behind Sivy. It’s the tool I wish my friend had.

Sivy is an AI-powered recruitment intelligence platform that transforms that chaotic pile of PDFs into a clean, ranked, and searchable talent pool. You upload the CVs, define a role profile with the technical and cultural requirements you're looking for, and the AI does the heavy lifting.

It provides a technical score, a culture fit score, a professional summary, and a list of extracted skills for every single candidate. The two-day manual process becomes a two-minute automated one.

The journey of building Sivy has been challenging, but it’s driven by the memory of that frustrating weekend. It’s about solving a real, tangible problem and hopefully, making the process of connecting great people with great opportunities just a little bit more intelligent and a lot more human.

This is just the beginning, and I’m excited to share more of the journey as we build.