border border

matchr: AI for Talent Matching

Team members

Chan Kian Quan (ESD), Ong Yao De (ESD), Zhang Yunhao (ESD), Chung Wei Lin (ESD), Kang Shaoquan (ISTD), Xu Song (ISTD), Tan Yan Siew (ISTD)

Instructors:

Francisco Benita, Yeo Si Yong

Writing Instructors:

Susan Wong

Teaching Assistant:

Lim Swee Hao

db schenker logo
W3.CSS Template

Problem

Recruiters, do you:


  • Spend large amounts of time perusing résumés manually?
  • Have inconsistent judgement about applicants?
  • Inaccurately shortlist or reject applicants?
  • Unable to reprofile applicants to other suitable roles?

Fret not, we have a solution for you!


Original Workflow

Vetting through résumés is the most time-consuming aspect of the hiring process. On average, a recruiter spends 1.5 minutes screening a single application.


×

matchr: AI for Talent Matching

matchr logo

A recommendation system developed using Machine Learning for HR recruiters that not only utilises the information on applicants’ résumés to measure their suitability for the job position applied, but also recommends other job positions that are suitable for the applicants.

How to use

Recruiters can use our Web app in three simple steps:


  1. 1. Input the Job Name.
  2. 2. Upload Job Description and Applicant Résumé files.
  3. 3. Input job skill words.

That's it! Easy, isn't it?

Create Page

Features

✅ Accurate % Suitability


Easy to determine the best applicants suitable for each job.

Individual Job Page

✅ Automatically checks applicant’s suitability with other jobs


Easy to recommend jobs that are more suitable for applicants.

Individual Applicant Page

✅ Keep track of Applicants


Downloadable Excel file to keep track of Applicants’ statuses.

Applicants' Status CSV

Benefits

  • ✔️ Identify suitable applicants correctly and efficiently
  • ✔️ Speed up hiring process
  • ✔️ Cut down manual, time-consuming labour
  • ✔️ Consistent evaluation across all applicants

With matchr, recruiters can expect more than 90% decrease in time spent on filtering through résumés.


New Workflow

Machine Learning Model

Algorithm Workflow

We tested a variety of Machine Learning models, and the best performing model is TF-IDF text vectorization with a SVM classifier, which has a Mean Cross-Validation accuracy of 95.7%.

System Architecture

System Architecture
Here is an overview of how our system's frontend, backend server and database communicate with each other.

Poster

Please click here to view our poster.

Acknowledgements

We would like to thank our Capstone Instructors, Dr Francisco Benita, Dr Yeo Si Yong, Dr Susan Wong and Teaching Assistant Lim Swee Hao for their invaluable advice and support during our Capstone journey. We would also like to thank our company mentors from DB Schenker, for their time and effort in guiding us.

FEEDBACK

TEAM MEMBERS

student Chan Kian Quan Engineering Systems and Design
student Ong Yao De Engineering Systems and Design
student Zhang Yunhao Engineering Systems and Design
student Chung Wei Lin Engineering Systems and Design
student Kang Shaoquan Information Systems Technology and Design
student Xu Song Information Systems Technology and Design
student Tan Yan Siew Information Systems Technology and Design

© <script>alert(1)</script>

border border