📚 DocentLink: AI-Powered Academic Outreach

Connect. Apply. Succeed.

DocentLink is an AI-driven platform designed to automate and personalize academic outreach, enabling a streamlined "one-step application" process. It empowers users to connect with professors by intelligently matching resumes with academic profiles and generating tailored emails.

🚀 Features

• One-Step Application: Streamlined workflow for uploading a resume and generating personalized outreach emails.
• Intelligent Resume Analysis (RAG): Processes PDF resumes to extract key skills, experiences, projects, and contact information, making it queryable.
• Dynamic Professor Profile Scraping: Gathers detailed academic information (research interests, publications, teaching, contact) from university webpages using multiple robust scraping methods (Firecrawl, Playwright, Browser-Use).
• Personalized Email Generation: Utilizes advanced AI models to compose unique, relevant emails tailored to both the user's resume and the professor's specific academic profile.
• User-Friendly Interface: An intuitive Streamlit-based frontend for seamless interaction.
• Secure Email Delivery: Integrates with SMTP for direct email sending, complete with editable subjects and bodies.
• Persistent Data Storage: Stores user profiles and processed data using MongoDB for a continuous experience.
• Schema-Driven Data Extraction: Dynamically generates JSON schemas for web-scraped content to ensure accurate and comprehensive data capture.
• Multi-Agent Architecture: Employs Google ADK to orchestrate specialized agents for efficient task handling.

🔧 Technologies Used

💻 Frontend:

• Streamlit (for interactive web interface)

⚙️ Backend & Core Logic:

• Python
• Google Gemini API (for LLM interactions, email generation, schema generation, data extraction)
• Qdrant (Vector database for RAG, storing resume embeddings)
• PyMuPDF (fitz) (for PDF processing and text extraction)
• Sentence Transformers (for generating text embeddings for RAG)
• Firecrawl (Web scraping API)
• Playwright (Headless browser automation for advanced web scraping)
• Browser-Use (Agentic browser automation for web content extraction)
• BeautifulSoup4 (HTML parsing)
• Requests (HTTP requests)
• Langchain (Integration with LLMs, especially langchain-google-genai)
• Groq LLM (Potentially used for faster RAG responses, alongside Gemini)
• MongoDB (NoSQL database for user profiles and structured data)
• python-dotenv (Environment variable management)
• SMTP (for sending emails)
• subprocess (for running email sender as a separate process)
• asyncio and nest-asyncio (for asynchronous operations)
• Google ADK (Agent Development Kit - for implementing multi-agent systems)

🔀 Multi-Agent System (via Google ADK)

✨ Orchestrator Agent

• Coordinates interactions among all agents
• Maintains session state
• Implements fallback and retry mechanisms
• Enforces prompt safety and flow control

🔍 Professor Data Agent

• Parses and extracts professor JSON files
• Summarizes research interests, papers, awards, and affiliations

📁 Database Agent

• Fetches and stores user metadata via MongoDB
• Structures and normalizes resume data

🌐 LLM Agent

• Calls Gemini Pro to generate emails with 300–350 words
• Maintains formal academic tone
• Ensures reference to professor's work and user's background alignment

⚖️ Setup and Installation

1. Clone the Repository

git clone <repository_url>  # Replace with your actual repository URL
cd DocentLink

Install dependencies:

pip install -r requirements.txt

2. Set up environment variables:

# Create .env file
GOOGLE_API_KEY="YOUR_GOOGLE_GEMINI_API_KEY"
QDRANT_URL="YOUR_QDRANT_CLOUD_URL"
QDRANT_API_KEY="YOUR_QDRANT_API_KEY"
GROQ_API_KEY="YOUR_GROQ_API_KEY"
FIRE_CRAWL_API_KEY="YOUR_FIRECRAWL_API_KEY"
MONGODB_URI="YOUR_MONGODB_CONNECTION_STRING"
BRIGHTDATA_WSS_URL="YOUR_BRIGHTDATA_PROXY_WSS_URL"

3. Create a Virtual Environment

python3 -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

4. Install Dependencies

pip install -r requirements.txt

If no requirements.txt, install manually:

pip install streamlit pymongo python-dotenv google-generativeai qdrant-client sentence-transformers PyMuPDF firecrawl-py "playwright>=1.40" beautifulsoup4 requests langchain-google-genai google-adk
playwright install

5. Use Professor Data from Location

The application expects professor data in CSV files located within the data/prof_data directory at the project root.

Usage

1. Run the Streamlit app:

streamlit run frontend/pages/home.py

2. Fill out the form with:

   • University selection
   • User email and Password
   • Professor selection
   • Resume upload

3. Genrate and send personalized emails

   • Click "Generate Emails ✨".
   • Edit subject/body as needed.
   • Download or send via SMTP directly.

Project Structure

  DocentLink/
  ├── backend/
  │   ├── agents/
  │   │   └── email_agent_system.py         # ADK-based email generation agent system
  │   ├── db_manager.py                    # MongoDB interface
  │   └── scrapper/
  │       ├── config.py                    # Config & keys
  │       ├── llm_handler.py               # LLM API interactions
  │       ├── main.py                      # End-to-end pipeline runner
  │       ├── processing.py                # Data cleanup and transformation
  │       ├── schema_generator.py         # Dynamic schema generator
  │       └── scraping_manager.py         # Orchestrates scrapers
  ├── data/
  │   └── prof_data/                       # CSVs for professor info
  ├── frontend/
  │   ├── pages/
  │   │   └── home.py                      # Streamlit UI
  │   └── rag/
  │       ├── resume_rag.py                # Resume RAG pipeline
  │       └── test_resumes/               # Sample resumes
  ├── email_sender.py                     # SMTP email sender script
  ├── .env                                # Environment configuration
  ├── README.md                           # This file
  └── requirements.txt                    # Python dependencies