PRAVEEN/YADAV
All projects
05 / Internal service · Search

Candidate Search & Relevance Engine

Search that ranks candidates for a role and shows its working.

Built solo
Backend & infrastructure
What it is

An internal search service that ranks candidates against a job requisition and shows recruiters how each score was worked out.

My role

Built solo

Category

Internal service · Search

Main stack

Node.js · Express · MongoDB · Mongoose

Highlight

Skills first, then the practical stuff

Status

Private internal service

ARCHITECTURE DIAGRAM
01Query
02Relevance
03Explanation
Diagram · how it is put together
WHY IT EXISTS

Search that ranks candidates for a role and shows its working.

Recruiters don’t want every CV that contains the word React. They want the people closest to one specific requisition, in order, with a reason. Plain keyword search answers the wrong question. This service sits beside an applicant tracking system, scores candidates against a role on five factors, and returns the breakdown with every result.

What I did

I designed and built this service on my own. It is internal, so there is no link and no real data here, just a small demo with made-up candidates.

5ranking factors, skills weighted highest
222skill synonym groups
3ways the index stays in sync
PROJECT OVERVIEW

How it fits together.

The product, in five steps.

01Open requisition
02Search candidates
03Rank by fit
04Show the reasoning
05Recruiter decides
WHAT I FOCUSED ON
01

Skills first, then the practical stuff

Skills carry most of the weight, then location, experience, salary and notice period. If a recruiter only specifies skills, the score is normalised against what they asked for, so a skills-only search doesn’t cap everyone at a low number and look broken.

02

Every score shows its breakdown

Each result lists which must-have skills matched, which didn’t, and how location, experience, salary and notice period came out. The point is that a recruiter can disagree with the ranking and see why.

03

It speaks recruiter

I built a vocabulary of 222 skill groups, so “dotnet” finds someone whose CV says C# and “node” finds Node.js. It is a curated list of aliases and related technologies for this domain, not a generic thesaurus.

TRY IT

See how the requirements change the order

Invented candidates. Nothing here is connected to a real applicant system.

Change the skill or the location preference and watch the ranking move.

Candidate Alpha

React · Node.js / Mumbai

75%Example match
Candidate Beta

React · Python / Pune

75%Example match
Candidate Gamma

Python · Node.js / Mumbai

0%Example match
BUILT WITH
Node.jsExpressMongoDBMongooseMeilisearch (self-hosted)node-cronWinstonHelmetPM2
WHERE IT STANDS

Private internal service. The demo here uses invented candidates and isn’t connected to anything real.

Numbers on this page are as of September 2026.
INTERESTED?

Want to hear more about my part of it?

I can’t share access to this one since it is company work, but I am glad to talk through the pieces I owned and the decisions behind them.

Ask for a walkthroughOpens your email client · ypraveen760@gmail.com