Job Autopilot 🚀

AI-powered job application automation system that streamlines your job search workflow using GPT-4o, automated scraping, resume optimization, LinkedIn outreach, and intelligent cold email campaigns.

Perfect for: Job seekers in EdTech, L&D, AI Product Management, and Automation fields

---

✨ Features

• 🔍 Automated Indeed Scraping via Apify
• 🤖 AI-Powered Job Scoring (0-10 rating based on your profile)
• 📊 Smart Categorization (EdTech, AI PM, Automation, L&D)
• 💾 Database Caching (Neon PostgreSQL + Local SQLite fallback)
• 📦 Load Cached Jobs (reuse previous searches, save API quota)

• 📤 Multi-Format Upload: Support PDF, DOCX, and Markdown master resumes
• 🎨 Professional Templates: 4 ATS-friendly templates (single/two-column, classic/modern)
• 🧠 GPT-4o Powered: Resume optimization uses GPT-4o for higher accuracy
• 📊 ATS Scoring: Real-time ATS compatibility score with keyword matching
• 🎯 Job-Tailored Resumes: AI optimizes resume for each job description
• 🔒 Anti-Hallucination: Iron-clad data protection - dates, locations locked

• 🎓 School Configuration: Set your alumni schools (priority-ranked)
• 🏷️ Target Fields: Define your professional interests (L&D, AI, etc.)
• 📋 Job Integration: Link high-value jobs to LinkedIn search
• 🔗 One-Click LinkedIn Launch: Search alumni directly from UI

• 🌐 Chrome DevTools MCP: AI-friendly browser automation via accessibility tree
• 🎓 Alumni Search: Find 2nd-degree connections from your schools
• 🤖 AI Agents:
  • ContactRankerAgent: Priority scoring (0-100) based on job match, alumni status
  • ScamDetectionAgent: Filter suspicious profiles automatically
  • PersonalizationAgent: GPT-4 powered message generation
• 🧠 Memory Layer: ChromaDB vector storage for learning from successful messages
• 📨 Auto-Connect: Send connection requests with rate limiting (10-20s delays)
• 💾 Persistent Profile: Uses dedicated Chrome profile (no conflict with personal Chrome)

• 📝 Draft Cold Emails: AI-generated personalized emails
• 📬 Gmail Integration: Create drafts directly in Gmail
• 📊 Email Statistics: Track drafts, sent, replied counts
• ⏰ Follow-up Queue: Auto-generate follow-up drafts

• 📈 Kanban Board: Visual pipeline (To Apply → Sent → Replied → Interview)
• 📚 Applied History: Manage manually marked applications
• 🎯 One-Click Apply: Move jobs through stages
• 📧 Status Tracking: Real-time application status

---

🏗️ Architecture

---

🛠️ Tech Stack

Category	Technology
AI/LLM	OpenAI GPT-4o / GPT-4o-mini
Frontend	Streamlit 1.30+
Backend	Python 3.11+
Database	Neon PostgreSQL (cloud)
Job Scraping	Apify (Indeed Actor)
LinkedIn Automation	Chrome DevTools MCP (Puppeteer-based)
Memory Layer	ChromaDB (Vector Database)
Email	Gmail API (OAuth 2.0)
Resume	python-docx, ReportLab (PDF)
ORM	SQLAlchemy 2.0

---

🚀 Quick Start

Prerequisites

• Python 3.11+
• Node.js 18+ (for Chrome DevTools MCP)
• Git

Installation

# 1. Clone the repository
git clone https://github.com/Schlaflied/job-autopilot.git
cd job-autopilot

# 2. Create virtual environment
python -m venv venv
venv\Scripts\activate  # Windows
source venv/bin/activate  # macOS/Linux

# 3. Install dependencies
pip install -r requirements.txt

# 4. Configure environment variables
cp .env.example .env
# Edit .env with your API keys

# 5. Initialize database
python scripts/init_database.py
python scripts/init_coffee_chat_db.py

# 6. Run the application
streamlit run streamlit_app.py --server.port=8502

Access the app: http://localhost:8502

🐳 Docker Deployment

You can run the entire application in a Docker container (recommended for stability).

# 1. Build the image
docker build -t job-autopilot .

# 2. Run the container
# We map port 8502 and allow access to the host's Chrome (CDP)
docker run -p 8502:8502 -p 5000:5000 \
  --add-host=host.docker.internal:host-gateway \
  --env-file .env \
  -v "%cd%/data":/app/data \
  job-autopilot

Note: The --add-host flag is crucial for the Docker container to connect to your local Chrome instance for LinkedIn automation.

---

📁 Project Structure

job-autopilot/
├── modules/
│   ├── ai_agent.py              # GPT-4o integration (scoring, resume, emails)
│   ├── coffee_chat_agents.py    # ✨ AI Agents (Ranker, Scam, Personalization)
│   ├── coffee_chat_memory.py    # ✨ ChromaDB Memory Layer
│   ├── coffee_chat_models.py    # SQLAlchemy models for Coffee Chat
│   ├── linkedin_automation.py   # ✨ LinkedIn search and automation
│   ├── job_scraper.py           # Apify job scraper with caching
│   ├── job_contact_integrator.py # Job + Contact integration
│   ├── gmail_service.py         # Gmail API integration
│   ├── database.py              # SQLAlchemy models (Neon PostgreSQL)
│   ├── resume_generator.py      # Resume PDF/DOCX generation
│   └── logger_config.py         # Centralized logging
├── pages/
│   ├── coffee_chat_center.py    # ✨ Coffee Chat Dashboard
│   └── user_profile.py          # ✨ School & Fields Configuration
├── scripts/
│   ├── linkedin_auto_connect.py # ✨ End-to-end LinkedIn automation
│   ├── init_database.py         # Database initialization
│   └── init_coffee_chat_db.py   # Coffee Chat tables
├── docs/
│   └── COFFEE_CHAT_PLAN/        # LinkedIn & Coffee Chat documentation
├── streamlit_app.py             # Main Streamlit UI
├── requirements.txt             # Python dependencies
└── README.md                    # This file

---

⚙️ Configuration

Required API Keys

1. OpenAI API (AI features)

OPENAI_API_KEY=sk-proj-your_openai_api_key_here
OPENAI_MODEL=gpt-4o-mini

• Get key: https://platform.openai.com/api-keys

2. Apify API (Job scraping)

APIFY_API_TOKEN=apify_api_your_token_here

• Get token: https://console.apify.com/account/integrations

3. Neon PostgreSQL (Database)

DATABASE_URL=postgresql://user:password@host.neon.tech/dbname?sslmode=require

• Get database: https://neon.tech/

4. Gmail API (Email automation)

GMAIL_CREDENTIALS_PATH=./data/credentials/gmail_credentials.json
GMAIL_TOKEN_PATH=./data/credentials/gmail_token.json

---

🤖 Browser Automation (MCP-Style)

This project uses a direct connection to Chrome's DevTools Protocol (CDP) to "read" web pages and perform actions like a human agent.

1. Launch Debug Chrome

To allow the app to control your browser, you must launch Chrome with a specific debugging port (9222).

Windows Users:

1. Locate launch_chrome_debug.bat in the project root.
2. Double-click to run it.
3. A new Chrome window will open. Log in to LinkedIn in this window.

2. Use Coffee Chat Center

1. Go to the Coffee Chat Center page in the app.
2. Click "🔗 Test Connection" to verify the app can "see" your Chrome window.
3. Use features like "Dive & Draft":
   • The app will navigate Chrome to a contact's profile.
   • It reads the screen (using MCP/CDP).
   • AI analyzes the profile hooks and drafts a message.
   • Click "Paste to LinkedIn" to insert the draft.

---

🔐 Security & Privacy

Your Data Stays With You.

• Local Database: Job data is stored in your local PostgreSQL/SQLite instance.
• Local Browser: Automation happens on your machine, in your browser.
• Git Ignoring: The .gitignore is pre-configured to exclude:
  • docs/ (Your personal strategies/logs)
  • data/ (Resumes, credentials)
  • chroma_data/ (AI memory)
  • *.csv (LinkedIn exports)

To Import LinkedIn Connections:

1. Export your data from LinkedIn Settings (Connections.csv).
2. Place it in the project folder.
3. Run the import script (or use the UI uploader).

   python scripts/import_enhanced_connections.py
   

---

Command Line Usage

# Direct script execution
python scripts/linkedin_auto_connect.py --company "google.com" --school "University of Western Ontario" --limit 5

---

💰 Cost Estimate

Service	Cost	Notes
OpenAI GPT-4o-mini	~$5-10/mo	Job scoring + resume + emails
OpenAI Embeddings	~$0.30/1000 contacts	Memory Layer vectors
Apify (Indeed scraper)	$0 (free tier)	$5 free credit
Neon PostgreSQL	$0 (free tier)	0.5GB storage
Gmail API	$0	Free for personal use
Total	$5-10/mo	Scalable to 100+ applications

---

🐛 Troubleshooting

"Chrome already running"

taskkill /F /IM chrome.exe

"LinkedIn not loading"

• The script uses a dedicated profile at C:/temp/linkedin-automation-profile
• First run requires manual LinkedIn login
• Login persists for future runs

"No 2nd degree connections found"

• Try different company/school combinations
• Some companies have few alumni in your network

---

📜 License

This project is licensed under GNU Affero General Public License v3.0 (AGPL-3.0).

• ✅ Free to use, modify, distribute
• ⚠️ Must open-source modifications under same license
• ⚠️ Network users entitled to source code

---

🙏 Acknowledgments

• OpenAI for GPT-4o API
• Google Chrome DevTools Team for Chrome DevTools MCP
• Apify for job scraping infrastructure
• Neon for free PostgreSQL tier
• ChromaDB for vector database
• Resume-Matcher for inspiring our PDF/DOCX parsing approach using pdfminer.six and docx2txt
• Job seekers worldwide 💪

---

🎯 Roadmap

• LinkedIn auto-connect with AI agents
• Memory layer for learning from successful messages
• Coffee chat message automation (post-connection)
• Multi-language support
• Interview prep AI coach

---

⭐ Star this repo if it helped you land a job! ⭐

Report Bug · Request Feature