🤖 Job Agent — AI-Powered Job Application System

A full-stack Agentic AI system for automated job searching, resume tailoring, and cover letter generation across multiple job platforms.

🧠 AI-Powered | Multi-Agent Pipeline | LangGraph Orchestrated
🌐 Searches LinkedIn, Remotive, Internshala simultaneously
📄 Tailors your resume and writes cover letters automatically

📸 Project Screenshots

🖥️ Live Dashboard

📌 About the Project

Job Agent is a multi-agent AI application that automates the job application process end-to-end. Instead of manually browsing job boards and writing cover letters, this system does it all — finds jobs, scores them against your resume, writes tailored documents, and tracks applications.

Built using LangGraph for agent orchestration, Groq LLM for AI inference, and FastAPI for the backend — all accessible through a sleek real-time web dashboard.

✨ Features

Feature	Description
🔍 Multi-Source Scraping	Searches LinkedIn, Remotive, Internshala at once
🧠 AI Job Scoring	Scores each job 0-100 against your resume
✍️ Cover Letter Generation	Writes a 250-300 word personalized cover letter
📄 Resume Tailoring	ATS-optimized resume with job-specific keywords
💡 AI Role Suggestions	Suggests 5 roles based on your uploaded resume
📊 Live Dashboard	Real-time streaming pipeline with job cards
🔗 One-Click Apply	Opens real job pages with apply button
📋 Application Log	Saves all applications to JSON + optional Notion sync
🗑️ Clear Log	Delete saved applications from dashboard

🛠️ Tech Stack

Layer	Technology
Agent Orchestration	LangGraph StateGraph
AI / LLM	Groq — Llama 3.3 70B (Free)
Backend	FastAPI + Uvicorn
Frontend	HTML / CSS / Vanilla JS
Resume Parsing	PyMuPDF (PDF), python-docx (DOCX)
Job Sources	Remotive API, LinkedIn, Internshala
Optional Sources	Adzuna API (Naukri + Indeed aggregated)
Browser Automation	Playwright (optional)
Tracking	Local JSON + Notion API (optional)
Environment	Python 3.11+, Virtual Environment

📁 Project Structure

job-agent/
├── agents/
│   ├── scraper.py        # Finds jobs from LinkedIn, Remotive, Internshala
│   ├── planner.py        # AI scores each job against your resume (0-100)
│   ├── critic.py         # Filters low scores, selects best match
│   ├── writer.py         # Writes tailored cover letter + resume
│   ├── browser.py        # Playwright browser automation (optional)
│   └── tracker.py        # Logs applications to JSON + Notion
│
├── core/
│   ├── state.py          # Shared AgentState TypedDict schema
│   ├── graph.py          # LangGraph orchestrator with routing
│   └── memory.py         # Application history and memory
│
├── tools/
│   ├── llm.py            # Groq API wrapper
│   └── browser_tools.py  # Playwright helper functions
│
├── static/
│   └── index.html        # Full web dashboard (single file)
│
├── data/
│   └── applications.json # Auto-generated application log
│
├── server.py             # FastAPI server with all endpoints
├── main.py               # CLI entry point (backup)
├── .env                  # Secret API keys — never commit this
├── .gitignore            # Ignores venv, .env, __pycache__
└── README.md             # This file

🔄 How It Works

 User uploads Resume (PDF/DOCX)
            │
            ▼
    ┌───────────────┐
    │  Scraper      │  →  Searches LinkedIn + Remotive + Internshala
    └──────┬────────┘
           │
           ▼
    ┌───────────────┐
    │  Planner      │  →  AI scores each job 0-100 vs your resume
    └──────┬────────┘
           │
           ▼
    ┌───────────────┐
    │  Critic       │  →  Approves jobs scoring 50+, skips duplicates
    └──────┬────────┘
           │
           ▼
    ┌───────────────┐
    │  Writer       │  →  Writes tailored resume + cover letter
    └──────┬────────┘
           │
           ▼
    ┌───────────────┐
    │  Tracker      │  →  Saves log to JSON + syncs to Notion
    └───────────────┘
           │
           ▼
    Dashboard shows job cards + generated documents
    User clicks "View & Apply →" to open real job page

Orchestrator: LangGraph StateGraph with conditional edges — routes to writer only if critic approves a job, otherwise exits cleanly.

⚙️ Installation & Setup

Prerequisites

Make sure these are installed on your machine:

✅ Python 3.11+
✅ VS Code
✅ Git

Step 1 — Clone the Repository

git clone https://github.com/yourusername/job-agent.git
cd job-agent

Step 2 — Create Virtual Environment

# Create virtual environment
python -m venv venv

# Activate — Windows
venv\Scripts\activate

# Activate — Mac / Linux
source venv/bin/activate

✅ You should see (venv) appear at the start of your terminal line

Step 3 — Install All Dependencies

pip install langgraph langchain-anthropic anthropic playwright \
            beautifulsoup4 requests notion-client python-dotenv \
            groq fastapi uvicorn python-multipart pymupdf python-docx

Install the browser engine:

playwright install chromium

Step 4 — Create `.env` File

Create a file named .env in the root job-agent/ folder and paste:

GROQ_API_KEY=gsk_your-groq-key-here
NOTION_API_KEY=secret_your-notion-key-here        # optional
NOTION_DATABASE_ID=your-database-id-here          # optional
ADZUNA_APP_ID=your-adzuna-app-id                  # optional
ADZUNA_APP_KEY=your-adzuna-app-key                # optional

⚠️ IMPORTANT: Never share this file. Never push it to GitHub. It is already listed in .gitignore.

Step 5 — Run the Server

uvicorn server:app --reload

Open your browser and go to:

http://127.0.0.1:8000

🖥️ Usage

1.  Open http://127.0.0.1:8000
2.  Upload your resume — drag & drop or click to browse (PDF, DOCX, TXT)
3.  AI analyzes resume → shows 5 suggested roles at the bottom
4.  Type your job search → e.g. "cybersecurity analyst"
5.  Click RUN AGENT
6.  Watch the live pipeline:
    scraper → planner → critic → writer → tracker
7.  Job cards appear with AI scores (green border = approved)
8.  Cover letter + tailored resume shown at bottom
9.  Click "View & Apply →" to open the real job page
10. Click "Copy" → paste your cover letter into the application

🤖 Agent Pipeline

🔍 Scraper Agent

Searches multiple job platforms with your exact query:

Remotive — remote tech jobs via free public API
LinkedIn — public job listings without login required
Internshala — internships and fresher jobs in India
Adzuna (optional) — aggregates Naukri + Indeed + more

🧠 Planner Agent

Sends each job + your resume to Groq LLM for scoring. Returns a score 0-100 and a one-line reason. Falls back to title-match scoring if job description is missing.

🔎 Critic Agent

Quality gate — approves jobs scoring 50 or above. Skips jobs already applied to. Picks the highest scoring job as the writing target.

✍️ Writer Agent

For the selected job, generates two documents:

Cover Letter — 250-300 words, personalized to the role and company
Tailored Resume — ATS-optimized with keywords from the job description

📋 Tracker Agent

Saves application details to data/applications.json with timestamp and fit score. Optionally syncs to a Notion database table.

🌐 Job Sources

Platform	Type	Method
Remotive	Remote tech jobs worldwide	Free public API
LinkedIn	All job types, India focus	Public guest search
Internshala	Internships + fresher India	Web scraping
Adzuna (optional)	Naukri + Indeed aggregated	Free API key required

🔑 API Keys Required

Groq API — Free, Required

Go to console.groq.com
Sign up free
Click API Keys → Create API Key
Copy the key starting with gsk_...
Paste into .env as GROQ_API_KEY

Adzuna API — Free, Optional (adds Naukri + Indeed)

Go to developer.adzuna.com
Sign up free → Create Application
Select Job Search, India as primary market
Copy App ID and App Key from dashboard
Paste into .env as ADZUNA_APP_ID and ADZUNA_APP_KEY

Notion API — Free, Optional (application tracking dashboard)

Go to notion.so/my-integrations
Create new integration → copy the Internal Integration Secret
Create a Table page in Notion → connect your integration via ... menu
Copy the database ID from the URL
Paste into .env as NOTION_API_KEY and NOTION_DATABASE_ID

🔄 Running Again Later

Every time you want to start the project:

# Step 1 — go to project folder
cd D:\job-agent

# Step 2 — activate virtual environment
venv\Scripts\activate          # Windows
source venv/bin/activate       # Mac/Linux

# Step 3 — start the server
uvicorn server:app --reload

# Step 4 — open browser
# Go to: http://127.0.0.1:8000

Press Ctrl + C in terminal to stop the server.

🛠️ Troubleshooting

venv\Scripts\activate fails on Windows:

Set-ExecutionPolicy -ExecutionPolicy RemoteSigned -Scope CurrentUser

uvicorn not recognized:

pip install uvicorn

API authentication error:

Check .env file exists in root folder
No spaces around = sign: KEY=value not KEY = value
No quotes around the key values

No jobs found:

Try a simpler search: python developer instead of very specific terms
Remotive always works — LinkedIn may block requests occasionally
Add Adzuna key for more consistent results

Suggestions not loading:

System uses automatic keyword fallback
Even without AI it reads your resume and suggests roles based on detected skills

🚀 Future Improvements

Gmail watcher agent — detects company replies automatically
Scheduled daily runs — cron job applies to new jobs every morning
More job sources — Wellfound, WorkIndia, Shine
Interview preparation tips generated per job role
Application status tracker — Applied → Interview → Offer → Rejected
Email notifications for new matching jobs
Docker deployment for one-command setup

🎯 Interview Talking Points

When explaining this project, highlight these three concepts:

1. Reflexion Pattern The Critic agent rejects low-fit jobs before the Writer wastes tokens — demonstrating self-evaluation in agentic systems.

2. Conditional Edge Routing LangGraph routes to Writer only when Critic approves — showing production-level agent design with clean exit paths.

3. Multi-Modal Pipeline Scraping + LLM scoring + document generation + live streaming UI — end-to-end system thinking beyond a simple chatbot.

👨‍💻 Author

Vedanth D Built as an Agentic AI portfolio project.

LinkedIn: linkedin.com/in/vedanth-d-73a685296

📄 License

Apache License 2.0 — free to use, modify and distribute with attribution.

See LICENSE for full details.

Name		Name	Last commit message	Last commit date
Latest commit History 22 Commits
agents		agents
core		core
data		data
static		static
tools		tools
LICENSE		LICENSE
README.md		README.md
main.py		main.py
server.py		server.py

Folders and files

Latest commit

History

Repository files navigation

🤖 Job Agent — AI-Powered Job Application System

📸 Project Screenshots

🖥️ Live Dashboard

📋 Table of Contents

📌 About the Project

✨ Features

🛠️ Tech Stack

📁 Project Structure

🔄 How It Works

⚙️ Installation & Setup

Prerequisites

Step 1 — Clone the Repository

Step 2 — Create Virtual Environment

Step 3 — Install All Dependencies

Step 4 — Create .env File

Step 5 — Run the Server

🖥️ Usage

🤖 Agent Pipeline

🔍 Scraper Agent

🧠 Planner Agent

🔎 Critic Agent

✍️ Writer Agent

📋 Tracker Agent

🌐 Job Sources

🔑 API Keys Required

Groq API — Free, Required

Adzuna API — Free, Optional (adds Naukri + Indeed)

Notion API — Free, Optional (application tracking dashboard)

🔄 Running Again Later

🛠️ Troubleshooting

🚀 Future Improvements

🎯 Interview Talking Points

👨‍💻 Author

📄 License

About

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Step 4 — Create `.env` File

Packages