Abstract

This publication presents a multi-agent AI system designed to help people with their job search. The Multi-Agent Job Hunter uses five different AI agents that work together to analyze resumes, research job markets, create improved CVs, and match people with suitable jobs. The system was built to solve the problem of inefficient job searching by automating many time-consuming tasks. This solution shows how different AI agents can work together to provide practical career help. The system is currently in development and has only been tested by the development team.

1. Introduction

1.1 What This System Does

This system helps job seekers by automating the most time-consuming parts of looking for work. Instead of manually reviewing resumes, researching job markets, and applying to positions one by one, users can upload their resume and get comprehensive career assistance through AI automation.

The system provides:

• Detailed analysis of your resume with suggestions for improvement
• Research on current job market trends and opportunities
• Creation of optimized CVs tailored for specific roles
• Matching analysis between your background and available jobs
• Strategic advice on how to improve your applications

1.2 The Problem We're Solving

Looking for a job today involves many repetitive and time-consuming tasks:

• Resume Problems: Most people don't know if their resume is good enough or what's wrong with it
• Market Research: It's hard to know what jobs are available and what skills employers want
• Generic Applications: Sending the same resume to every job rarely works well
• Poor Matching: People often apply to jobs they're not well-suited for
• No Strategy: Most job seekers don't have a clear plan or professional guidance

Many resumes get rejected by computer systems before any human sees them. People spend hours each week on job searching without good results. The average job search takes several months, which shows there's room for improvement.

1.3 Our Solution

The Multi-Agent Job Hunter solves these problems by using five specialized AI agents that each handle different parts of the job search:

1. Coordinator Agent: Manages the workflow and decides which agent should work next
2. Resume Analyst: Reviews resumes and suggests improvements
3. Job Researcher: Finds job opportunities and analyzes market trends
4. CV Creator: Makes improved, professional CVs
5. Job Matcher: Compares your background with specific job requirements

The system connects to job search websites to get current job listings and creates a web interface where users can easily interact with all these features.

2. How the System Works

2.1 Overall Design

The system uses a central coordinator that manages five specialized agents. Think of it like a team where each person has a specific job, and there's a manager who decides who should work on what and when.

The Core Components:

Shared Information System:
The system keeps track of all information in one place so every agent can see what others have done. This includes the user's request, their resume, analysis results, job market data, and any files created.

Agent Management:
The Coordinator Agent acts like a project manager, deciding which agent should work next based on what the user wants and what information is already available.

2.2 The Five AI Agents

Coordinator Agent

• What it does: Decides which agent should work next and manages the overall process
• Why it's useful: Makes sure everything happens in the right order and no steps are skipped
• Technology: Uses GPT-4 to understand user requests and plan the workflow

Resume Analyst Agent

• What it does: Reviews resumes like an experienced HR manager would
• What it checks: Skills, experience, formatting, keywords, and how well it matches current job requirements
• Output: A detailed report with a score and specific suggestions for improvement

Job Researcher Agent

• What it does: Searches for jobs and analyzes what employers are looking for
• Data sources: Gets information from job posting websites
• Analysis: Looks at salary ranges, popular locations, required skills, and hiring trends

CV Creator Agent

• What it does: Creates a new, improved version of your resume
• Features: Optimized for computer screening systems, includes relevant keywords, professional formatting
• Technology: Creates PDF files that look professional and work well with job application systems

Job Matcher Agent

• What it does: Compares your background with specific job postings
• Matching process: Analyzes how well your skills and experience match job requirements
• Output: Gives you a score for each job and advice on how to improve your chances

2.3 Technical Setup

Technology Used:

• Programming: Python for the main system
• AI Models: OpenAI's GPT-4 for understanding and generating text
• Document Handling: Tools to read PDF and Word documents
• Web Interface: Flask framework for the website
• Hosting: Vercel cloud platform for reliable access
• Job Data: APIs from job search websites

Key Features:

• Handles Multiple Formats: Works with PDF, Word, and text files
• Error Handling: Continues working even when some parts have problems
• Keeps Track of Progress: Remembers what's been done and what's next
• Scalable: Can handle multiple users at the same time

3. System Workflow

3.1 How Users Interact with the System

The process is designed to be simple and straightforward:

Step 1: User Request

• User uploads their resume and describes what they want help with
• System analyzes the request to understand what services are needed

Step 2: Automatic Planning

• Coordinator Agent creates a plan for which agents to use and in what order
• System handles dependencies automatically (for example, running resume analysis before creating an improved CV)

Step 3: Agent Execution

• Each agent completes its specialized task
• Results are shared with other agents that might need the information
• Progress is tracked and reported to the user

Step 4: Results Delivery

• User receives comprehensive results including analysis, recommendations, and any generated files
• All information is presented in an easy-to-understand format

3.2 Workflow Diagram

3.3 Smart Coordination

The system is designed to be intelligent about the order of operations:

• If someone wants a new CV, the system first analyzes their current resume to understand what improvements are needed
• If someone wants job matching, the system first researches the job market to have current opportunities to compare against
• The Coordinator Agent automatically figures out these dependencies so users don't have to worry about the order of steps

4. Development Process

4.1 How We Built the System

We built the system in stages to make sure each part worked well before moving to the next:

Stage 1: Individual Agents

• Built each agent separately with its own specific capabilities
• Tested each agent to make sure it worked correctly
• Connected to external job search websites

Stage 2: Coordination System

• Created the Coordinator Agent to manage workflow
• Built the system for sharing information between agents
• Developed the logic for deciding which agent should work next

Stage 3: Integration and Testing

• Connected all the agents to work together
• Improved performance and fixed problems
• Built the web interface for users

Stage 4: Deployment

• Put the system online using cloud hosting
• Tested with real resumes and job searches
• Monitored performance and made improvements

4.2 Key Design Decisions

How Agents Communicate

All agents share information through a central system. This ensures everyone has access to the same data and prevents confusion or conflicts.

Handling Dependencies

The system automatically figures out when one agent needs information from another. For example, if the CV Creator needs resume analysis data that doesn't exist yet, it will automatically trigger the Resume Analyst first.

4.3 External Connections

Job Search Websites

• RapidAPI: Main source for current job postings - recommended
• ScraperAPI: Backup source for Google Jobs data - works well
• Adzuna API: Additional job market information - is a bit buggy, they seem to not be working

Document Processing

• Multiple Formats: Can read PDF, Word, and text files
• Text Extraction: Gets the important information from documents
• Quality Checks: Makes sure documents are readable and complete

5. Current Status and Testing

5.1 Development Team Testing

The system has been tested by our development team to ensure basic functionality works correctly:

What We've Tested:

• Resume Processing: Successfully reads and analyzes different resume formats
• Job Research: Retrieves current job postings from multiple sources
• CV Generation: Creates professional-looking PDF documents - this is not yet top notch. There are expected improvements to be made.
• Job Matching: Compares resumes against job requirements
• Overall Workflow: Complete process from upload to final results

Technical Performance:

To view the system performance report, you can check Career Assistant performance

• Processing Speed: Most operations complete within 2-3 minutes
• Error Handling: System continues working when external sites have problems
• File Handling: Successfully processes various resume formats
• System Stability: Runs reliably during development testing

5.2 Current Limitations

Since this is still in development, there are some important limitations:

Testing Status:

• Limited Testing: Only tested by the development team so far
• No User Studies: Haven't yet tested with real job seekers
• Performance Data: Don't have data on real-world effectiveness yet
• Feedback: Haven't collected user feedback on usefulness or accuracy

Technical Constraints:

• Processing Time: Takes several minutes to complete full analysis
• External Dependencies: Relies on job search websites being available
• Format Support: Only works with common document formats
• Language: Currently designed for English-language resumes only

5.3 What We're Planning to Test

User Testing:

• Real Job Seekers: Plan to test with people actually looking for work
• Different Industries: Test across various job types and experience levels
• Usability: Make sure the interface is easy to understand and use
• Effectiveness: Measure whether the system actually helps people get better results

Technical Validation:

• Load Testing: See how the system performs with multiple users
• Accuracy Testing: Verify that analysis and recommendations are helpful
• Reliability Testing: Ensure the system works consistently over time
• Security Testing: Make sure user data is protected

6. How to Use the System

6.1 Getting Started

To run this system locally on your computer, you can refer to the setup guide on the repository here :
Career Assistant Setup

else you can use this system via the web link where it has been deployed.

Using the system is designed to be straightforward:

Step 1: Access the Website

• Visit the web interface (hosted on Vercel) - Job-hunting-agent
• No account creation or login required for use

Step 2: Upload Your Resume

• Click the upload button and select your resume file
• System accepts PDF, Word, or text files
• File is processed securely and not stored permanently

Step 3: Describe What You Want

• Tell the system what kind of help you need. You can use one of our sample prompts
• Examples: "analyze my resume," "find software engineering jobs," "create an improved CV"
• System will understand and plan the appropriate workflow

Step 4: Wait for Processing

• System will show progress as different agents work
• Typical processing time is 2-3 minutes for complete analysis
• You can see which agent is currently working

Step 5: Review Results

• Get detailed feedback and recommendations
• Download any generated files (like improved CVs)
• All results are presented in easy-to-understand language

6.2 Types of Help Available

Resume Analysis

• Detailed review of your current resume
• Scoring based on industry standards
• Specific suggestions for improvement
• Identification of missing skills or experience

Job Market Research

• Current job opportunities in your field
• Salary information and location data
• Popular skills employers are looking for
• Market trends and demand levels

CV Creation - still in development

• Professional, optimized version of your resume
• Formatted to work well with automated screening systems
• Includes relevant keywords for your target jobs
• Available as a downloadable PDF

Job Matching

• Analysis of how well you match specific job postings
• Scoring for different opportunities
• Advice on improving your chances
• Strategy recommendations for applications

6.3 Best Results Tips

To get the most from the system:

• Use a Complete Resume: Include all relevant experience and skills
• Be Specific: Tell the system exactly what type of jobs you're interested in
• Provide Context: Mention your career goals or any specific challenges
• Check All Results: Review both the analysis and any generated documents
• Follow Recommendations: The system provides specific, actionable advice

7. Technical Implementation

7.1 System Architecture

The system is built using modern software development practices:

Core Framework:

class JobHuntingMultiAgent:
    def __init__(self):
        self.system = create_multi_agent_system()
    
    def process_request(self, user_message: str, resume_path: str = None):
        # Process user request through coordinated agents
        # Return comprehensive results

Agent Design:

Each agent follows the same basic pattern:

1. Check what information is available
2. Do its specific task
3. Update the shared information
4. Decide what should happen next
5. Report completion

7.2 Making It Work Reliably

Handling Problems:

• Backup Plans: If one job search site doesn't work, try others
• Error Messages: Clear explanations when something goes wrong
• Graceful Degradation: System keeps working even with partial failures
• Progress Tracking: Users always know what's happening

Performance:

To view the system performance report, you can check Career Assistant performance

• Efficient Processing: Agents work in the optimal order to save time
• Smart Caching: Avoid repeating the same work
• Resource Management: Use computing power efficiently
• Quality Checks: Verify that results are complete and useful

7.3 Security and Privacy

Data Protection:

• Local Processing: Resume content stays on our servers only during processing
• Temporary Storage: Files are automatically deleted after processing
• Secure Connections: All data transfer uses encryption
• No Long-term Storage: We don't keep your personal information

User Privacy:

• No Registration Required: Use the system without creating accounts
• Anonymous Processing: Your identity isn't required for the service
• Data Minimization: Only collect information needed for processing
• Transparent Handling: Clear information about how data is used

8. Future Plans

8.1 Immediate Improvements

User Testing and Feedback:

• Real User Testing: Test with actual job seekers to validate usefulness
• Interface Improvements: Make the system easier to use based on feedback
• Performance Optimization: Speed up processing and improve reliability
• Error Reduction: Fix problems discovered during broader testing

Feature Enhancements:

• More File Formats: Support for LinkedIn profiles and other resume formats
• Better Job Sources: Add more job search websites for broader coverage
• Improved Analysis: Make resume and job matching analysis more accurate
• User Guidance: Add more help and explanation throughout the process

8.2 Medium-term Goals

Enhanced Capabilities:

• Interview Preparation: Help users prepare for specific job interviews
• Salary Information: Better data on compensation for different roles
• Skills Gap Analysis: Detailed identification of skills to develop
• Application Tracking: Help users manage their job applications

Technical Improvements:

• Faster Processing: Reduce the time needed for analysis
• Better Integration: Connect with more career-related websites and tools
• Offline Capabilities: Some features that work without internet connection

8.3 Long-term Vision

Advanced Features:

• Career Planning: Long-term career development advice
• Industry Insights: Deep analysis of specific industries and trends
• Networking Help: Assistance with professional networking
• Continuous Learning: Recommendations for skills development

Broader Applications:

• Educational Use: Help students prepare for job searches
• Career Services: Tools for career counselors and HR professionals
• Company Integration: Help employers better match with candidates
• International Markets: Support for job searches in different countries

9. Limitations and Honest Assessment

9.1 Current Limitations

Testing Status:

• Early Stage: Only tested by developers, not real users yet
• No Effectiveness Data: Don't know yet if it actually helps people get jobs
• Limited Feedback: Haven't received input from actual job seekers
• Unproven Results: Can't claim improved job search outcomes yet

Technical Constraints:

• Processing Time: Takes several minutes, which might feel slow to users
• Internet Required: Needs constant internet connection to work
• External Dependencies: Performance depends on job search websites working
• Language Limitation: Only works well with English-language content

Scope Boundaries:

• No Interview Training: Doesn't help with actual interview skills
• Limited Personalization: Doesn't learn user preferences over time
• No Long-term Tracking: Doesn't follow up on job search success
• Generic Advice: Recommendations may not fit everyone's specific situation

9.2 What We Don't Know Yet

Effectiveness Questions:

• Does It Actually Help?: We don't know if users get better job search results
• User Satisfaction: Haven't measured if people find it useful
• Accuracy: Don't know how accurate the analysis and recommendations are
• Competitive Advantage: Uncertain if it's better than existing tools

Real-world Performance:

• Scalability: Don't know how it performs with many simultaneous users
• Reliability: Haven't tested long-term stability under real conditions
• User Behavior: Don't understand how people will actually use the system
• Edge Cases: May not handle unusual resumes or requests well

9.3 Plans for Validation

User Testing Strategy:

• Small Group Testing: Start with a small group of real job seekers
• Feedback Collection: Systematic collection of user experiences and suggestions
• Effectiveness Measurement: Track whether users actually get better results
• Iterative Improvement: Use feedback to make continuous improvements

Technical Validation:

• Performance Testing: Test with realistic user loads
• Accuracy Evaluation: Compare system recommendations with expert opinions
• Reliability Testing: Long-term monitoring of system stability
• Security Audit: Professional review of data protection measures

10. Conclusion

10.1 What We've Built

The Multi-Agent Job Hunter represents our attempt to make job searching more efficient and effective through AI automation. We've created a system where five specialized AI agents work together to provide comprehensive career assistance:

Technical Achievement: We successfully built a working system that coordinates multiple AI agents to handle different aspects of job searching. The agents communicate effectively and handle complex workflows automatically.

Practical Application: The system addresses real problems that job seekers face - poor resume quality, lack of market knowledge, and inefficient application processes.

User-Focused Design: We designed the system to be easy to use, with a simple web interface and clear results that don't require technical knowledge.

10.2 Current Status

What's Working:

• Basic Functionality: All core features work as designed
• Agent Coordination: The multi-agent system successfully manages complex workflows
• Document Processing: Successfully handles different resume formats
• Job Data Integration: Retrieves and analyzes current job market information
• File Generation: Creates professional-quality CV documents

What Needs Work:

• User Testing: We need to test with real job seekers to validate usefulness
• Performance Optimization: Processing time could be faster
• Error Handling: Some edge cases still need better handling
• User Experience: Interface could be more intuitive based on actual user feedback

10.3 Honest Assessment

This system is a working prototype that demonstrates the potential of multi-agent AI for career assistance. However, we're being honest about its current limitations:

It's Untested with Real Users: We've only tested internally, so we don't know yet if it actually helps people get jobs or if users find it valuable.

Technical Challenges Remain: The system works but has room for improvement in speed, reliability, and handling edge cases.

Market Validation Needed: We need to prove that this approach is better than existing solutions and worth the complexity.

10.4 Why This Matters

Despite the limitations, this project demonstrates several important concepts:

Multi-Agent Coordination: Shows how different AI systems can work together effectively on complex tasks.

Practical AI Application: Demonstrates AI solving real-world problems rather than just theoretical challenges.

Scalable Architecture: Provides a foundation that can be improved and expanded based on user feedback and needs.

Open Development: Documents the process and challenges for others working on similar problems.

10.5 Next Steps

Our immediate focus is on validation and improvement:

1. User Testing: Get the system in front of real job seekers and collect honest feedback
2. Performance Improvement: Make the system faster and more reliable
3. Feature Refinement: Improve existing capabilities based on user needs
4. Market Validation: Determine if this approach provides genuine value

The goal is to move from a working prototype to a genuinely useful tool that helps people navigate their careers more effectively. We believe the multi-agent approach has potential, but we need real-world testing to prove it.

---

Contact Information

For questions about this project, requests for user support or to participate in user testing:

Email: [amaechiisaac450@gmail.com]
Project Repository: [https://github.com/Dprof-in-tech/job-hunting-agent]

---

• The technical part of this system is licensed under the MIT license.

This publication documents an experimental multi-agent AI system for career assistance. The system is currently in development and has not yet been tested with real users. All claims about functionality are based on development team testing only.

💡 Liked this project? Show your support!

⭐ Give the project a star
🤝 Send me a Twitter / X connection request to stay in touch

Happy automation! 🚀📈