Job-Seeking Multi-Agent AI Assistant

This project showcases a multi-agent AI system built with Google’s Agent Development Kit (ADK) and Gemini models to automate and enhance the job application process.

Instead of relying on a single chatbot, this system uses multiple specialised AI agents, each responsible for a different task in the job-seeking lifecycle:

• Job and ATS analysis

• CV rewriting and optimisation

• Cover letter generation

• Interview coaching

• Similar job discovery (via Google search)

The result is a modular, reusable, and intelligent assistant that converts a job ad + CV into a complete, personalised application package.

---

1. The Problem

Job hunting today is not just about finding a role — it is about strategy, alignment, and precision. Many candidates struggle with:

• Understanding ATS (Applicant Tracking System) requirements
• Identifying key skills from job advertisements
• Tailoring CVs effectively
• Writing targeted cover letters
• Preparing job-specific interview answers
• Discovering similar opportunities efficiently

These steps are repetitive, time-consuming, and depend heavily on experience and insider knowledge.

This problem is especially significant for:

• Career switchers
• International applicants
• Early-career professionals
• Busy job seekers applying to multiple roles

---

2. The Solution

This project introduces a Multi-Agent Job Application Assistant that uses agentic AI to automate and improve the job-seeking workflow.

Instead of a single AI model, the system is divided into specialised agents:

• Agent Name Responsibility
• job_analyst_agent Extracts skills & generates ATS match score
• cvwriter_agent Rewrites CV to match job description
• clwriter_agent Creates a tailored cover letter
• coach_agent Generates interview questions + model answers
• research_agent Finds similar real-world job postings

Each agent has its own role, memory state, and output, and they are coordinated by an orchestration layer.

---

3. Architecture Overview

Below is the system workflow diagram showing how data moves through the multi-agent architecture:

---

4. How to Run the System

There are two files in the folder.

For the notebook:

Step 1: Set up your API key (securely)

If using Kaggle: Kaggle → Add-ons → Secrets → New Secret Name: GOOGLE_API_KEY Value: [your key]

Your notebook retrieves it securely with:

from kaggle_secrets import UserSecretsClient GOOGLE_API_KEY = UserSecretsClient().get_secret("GOOGLE_API_KEY")

Step 2: Run the notebook

You can run it in: Kaggle Notebook, Jupyter Notebook, or Google Colab

Step 3: Interactive mode

You’ll be prompted:

• Paste your job ad
• Paste current cv
• choose output type: cv, cover_letter, interview, jobs, all

The agents will then execute automatically and return results.

For the app file

The app.py file is a Streamlit-based interface for the Multi-Agent Job Application Assistant. It allows users to interact with the AI system through a simple web UI instead of a notebook.

The app connects to your agents and lets users paste a job description and a CV, then generate tailored outputs such as a revised CV, cover letter, interview guide, and similar job listings.

• Step 1 – Enter your Job Description
• Step 2 – Enter your Current CV
• Step 3 – Choose what to generate: Revised CV, Cover letter, Interview Guide, Similar Jobs, All
• Step 4 - Click the "Generate" button.
• Step 5 – Your results will appear in the app under clearly labelled sections.

---

5. Outputs Generated

The system can generate:

• ✅ ATS keywords + match %
• ✅ Rewritten, role-specific CV
• ✅ Professional cover letter
• ✅ Interview preparation guide
• ✅ Similar live job links

---

6. Future Enhancements

Potential improvements:

• LinkedIn integration
• Dashboard analytics
• Resume PDF export
• Multilingual support
• Better research abilities
• Long-term memory (Database for storage of job application and resumes)