Introduction

The growing popularity of computer science has led to an increasingly competitive job market, making the hiring process more challenging for both employers and developers. Employers struggle to identify the right talent from a vast pool of candidates, while many skilled developers find it difficult to secure job opportunities that match their expertise. DevX addresses these challenges by leveraging automated skill matching powered by Natural Language Processing (NLP) techniques such as BERT and cosine similarity. By analyzing resumes and job descriptions with advanced AI algorithms, DevX aims to bridge the connection between employers and developers.
The key objectives of DevX are:
• Address the challenges faced while hiring a developer.
• Address the challenges faced by a developer.
• Explore the possible method of solving/ automating the process of hiring a developer.
• Develop a platform to bridge employers and developers
This paper delves into the methodology, features and the implementation of DevX, including its future enhancements.

Features

Following are the key features of DevX that contributes to its functionality:
• User Registration and Authentication: Registration and login functionality with session management. Password hashing for security.
• Resume-Based Job Recommendation System: Matches developer profiles with job postings and recommends jobs based on skill similarity scores.
• Job Management System: Clients can create, manage and delete job postings, and use search functionality to find skilled developers. Developers can browse, filter and apply for relevant jobs.
• Review Management System: Clients can review and rate developers based on their performance.
• Chat System: Real-time messaging between developers and clients.

Database Collection and Description

The data for model training was scraped through the internet and compiled into one huge dataset.

Following is the drive link to the dataset:
https://drive.google.com/drive/folders/1nFfU4KnTuG0qwLD7myG9-3DqbFJTmaKi

Methodology

System Architecture

DevX adopts a three-tier architecture, where each component handles specific functionalities, creating a seamless user interaction, job postings and AI-driven skill matching.
Following are the key components used to build DevX:
• Front-end (React.js + Tailwind CSS): React.js was used to build a dynamic and seamless user interface, responsive across all devices. And Tailwind CSS was used for styling and building the UI components.
• Backend (Node.js + Express.js): This tech stack was used for processing user requests, storing data and interacting with the AI model.
• Database & AI Model (MongoDB + NLP-based AI Model): MongoDB was used for storing data and NLP-based AI Model was used to compare job descriptions with developer profiles.

Authentication & Security

• JSON Web Tokens (JWT): For secure user authentication and session management.
• bcrypt: For hashing passwords securely.

Process of Automated Skill Matching

1. Resume upload and Parsing: Developers upload their resume in any format. If the resume is text-based, the system extracts raw text directly, likewise, if it is in image format, the system utilizes OCR (Optical Character Recognition) to convert the image into machine-readable text. The system then undergoes cleaning and normalization. Any unwanted characters or stopwords are removed and the text is split into structured sections.
2. Skill Extraction and Categorization: The system identifies skills using an NLP technique named BERT- based Named Entity Recognition (NER). The model is pretrained on large datasets to identify common developer skills and new emerging technologies. After identifying the skills, the identified skills are grouped into predefined categories. Additionally, the system infers the developer’s role based on skill distribution and past experience.
3. Job Posting Analysis: After the client post job descriptions, the system extracts relevant text sections for analysis. Then, NLP techniques are used to identify soft vs hard skills, experience requirement and job type classification. The system then uses BERT embeddings to convert job descriptions into numerical vectors and use TF-IDF (Term Frequency-Inverse Document Frequency) for keyword importance.
4. Skill Matching and cosine similarity calculation: Both the Job postings and developer profiles are converted into numerical vectors using BERT embeddings and the cosine similarity between them is calculated. The closer the score is to 1, the better the match. Using the similarity score, jobs are ranked by relevance, with higher similarity scores appearing first.
5. Job Recommendation: The top N job postings are with the highest similarity scores selected and then displayed to the developer.

Result

While training the model, the model achieved an average accuracy of 87%

Future Prospects

• Interview Prep: Developers could be provided with interview preparations from industry experts on premium models.
• Expanded Skillset: The platform could be expanded to more skillset rather than just developers.
• AI assistant: Developers can be assisted with AI to make the perfect cover letter and assess their cv for better placement.

Conclusion

DevX is a freelance platform that leverages the power of AI to bridge the gap between developers and clients. Through its secure authentication system, resume-based job recommendations, real-time chats and user-friendly interface, the platform aims to offer a seamless and efficient recruitment experience for both clients and developers, essentially representing a next-generation hiring solution. Furthermore, the platform can be further scaled to using block-chain based contracts.

GitHub Link:

https://github.com/Aayush-prog/ReadyTensorApplication

References

Anna Stepanova, A. W. J. L. G. A. T. H., 2021. Hiring CS Graduates: What We Learned from Employers. ACM Transactions on Computing Education.
Ronak Surve, N. M. S. S. S. S., 2024. Job Analista : A Smart Resume Analyser and. International Journal of Advanced Research in Science, Communication and Technology (IJARSCT).
Vaswani, A. N. M. S. N. P. J. U. L. J. A. N. G. L. K. I. P., 2017. Attention is all you need.
s.l., Neural Information Processing Systems.