AI-Driven Resume Screening System with Chatbot Integration

An intelligent recruitment platform that automates end-to-end resume screening using AI/ML, delivering reliable candidate rankings, transparent explanations, and a seamless interview-preparation experience via an interactive chatbot. The system combines robust document parsing (PDF/DOCX/OCR), semantic matching with Sentence‑Transformers, fuzzy skill extraction, and fairness checks to help HR teams shortlist candidates faster while reducing bias and improving hiring quality.

Designed for scalability and extensibility, the project exposes REST APIs for integrations, supports role-based dashboards for HR and candidates, and includes automated notifications and explainability features (LIME + score breakdowns) so decisions remain auditable. Built as an academic capstone, it demonstrates production-ready patterns (JWT auth, async FastAPI, modular services) and serves as a foundation for future enhancements like S3 storage, interview scheduling, and multi-language support.

🎯 Overview

The AI-Driven Resume Screening System is a comprehensive web application designed to streamline the recruitment process for HR professionals while enhancing the candidate experience. Built as a major project at CMR Institute of Technology, Bengaluru, this system leverages cutting-edge AI/ML technologies to automate resume screening, provide intelligent rankings, and offer interview preparation assistance.

Key Capabilities

Automated Resume Parsing: Supports PDF, DOCX, PNG, and JPEG formats
AI-Powered Ranking: Multi-factor scoring using semantic similarity and skill matching
Explainable AI: Transparent decision-making with score breakdowns
Bias Detection: Fairlearn integration for fair hiring practices
Interview Chatbot: Rasa-powered assistant for candidate preparation
Email Notifications: Automated candidate updates
Dual Dashboards: Separate interfaces for HR and candidates

✨ Features

For HR Professionals

📝 Job Posting Management: Create and manage job descriptions with requirements
📤 Bulk Resume Upload: Process multiple resumes simultaneously (PDF/DOCX/Images)
🤖 AI-Powered Ranking: Automatic candidate scoring based on:
- Skill matching (45% weight)
- Semantic similarity (30% weight)
- Experience relevance (20% weight)
- Education level (5% weight)
📊 Explainable AI: Understand why each candidate received their score
✅ Decision Workflow: Select, reject, or mark candidates as pending
📧 Email Notifications: Automated emails sent to candidates upon decision submission
📈 Bias Detection: Fairlearn metrics to ensure fair evaluation

For Candidates

📱 Application Tracking: View status of all job applications
💬 Interview Prep Chatbot: 24/7 AI assistant offering:
- Common interview questions practice
- STAR method guidance for behavioral questions
- Technical interview tips
- Role-specific advice
🔔 Notifications: Real-time updates on application status
📊 Score Visibility: See ranking scores and feedback

🏗️ Architecture

graph TB
    subgraph "Frontend Layer"
        A[React Application<br/>Port 3000]
        A1[Landing Page]
        A2[Login/Register]
        A3[HR Dashboard]
        A4[Candidate Dashboard]
        A --> A1
        A --> A2
        A --> A3
        A --> A4
    end

    subgraph "Backend Layer"
        B[FastAPI Server<br/>Port 8000]
        B1[Authentication<br/>JWT + Supabase]
        B2[Job Management<br/>CRUD Operations]
        B3[AI Processor<br/>Resume Analysis]
        B4[Email Service<br/>Gmail SMTP]
        B --> B1
        B --> B2
        B --> B3
        B --> B4
    end

    subgraph "AI/ML Layer"
        C1[Text Extraction<br/>pdfplumber, pytesseract]
        C2[Skill Extraction<br/>spaCy, rapidfuzz]
        C3[Semantic Ranking<br/>Sentence-BERT]
        C4[Explainability<br/>LIME]
        C5[Bias Detection<br/>Fairlearn]
        B3 --> C1
        B3 --> C2
        B3 --> C3
        B3 --> C4
        B3 --> C5
    end

    subgraph "Chatbot Layer"
        D[Rasa Server<br/>Port 5005]
        D1[NLU Pipeline<br/>DIET Classifier]
        D2[Dialogue Management<br/>TED Policy]
        D3[Action Server<br/>Port 5055]
        D4[Custom Actions<br/>Interview Questions]
        D --> D1
        D --> D2
        D --> D3
        D3 --> D4
    end

    subgraph "Database Layer"
        E[(Supabase PostgreSQL)]
        E1[user_profiles]
        E2[job_descriptions]
        E3[resumes]
        E4[notifications]
        E --> E1
        E --> E2
        E --> E3
        E --> E4
    end

    subgraph "External Services"
        F[Gmail SMTP<br/>Email Delivery]
    end

    A -->|HTTP/REST| B
    A -->|WebSocket| D
    B -->|SQL| E
    B4 -->|SMTP| F
    D3 -->|Python| D4

    style A fill:#61dafb,stroke:#333,stroke-width:2px,color:#000
    style B fill:#009688,stroke:#333,stroke-width:2px,color:#fff
    style D fill:#5a67d8,stroke:#333,stroke-width:2px,color:#fff
    style E fill:#3ecf8e,stroke:#333,stroke-width:2px,color:#000
    style C3 fill:#ff9800,stroke:#333,stroke-width:2px,color:#000

Data Flow

sequenceDiagram
    participant HR as HR User
    participant FE as React Frontend
    participant BE as FastAPI Backend
    participant AI as AI Processor
    participant DB as Supabase DB
    participant Email as Gmail SMTP
    participant Candidate as Candidate User

    HR->>FE: Upload Resumes
    FE->>BE: POST /hr/jobs/{id}/upload-resumes
    BE->>AI: Extract Text (PDF/DOCX/OCR)
    AI->>AI: Extract Skills & Experience
    AI->>AI: Calculate Ranking Score
    AI-->>BE: Structured Data + Score
    BE->>DB: Store Resume Data
    DB-->>BE: Confirmation
    BE-->>FE: Ranked Candidates List

    HR->>FE: Make Decisions (Select/Reject)
    FE->>BE: POST /decisions/{resume_id}
    BE->>DB: Save Decision
    
    HR->>FE: Click "Submit Decisions"
    FE->>BE: POST /hr/jobs/{id}/submit-decisions
    BE->>DB: Fetch Candidates with Decisions
    DB-->>BE: Candidate Data
    
    loop For Each Candidate
        BE->>Email: Send Decision Email
        Email-->>Candidate: Email Notification
        BE->>DB: Create In-App Notification
    end
    
    BE-->>FE: Success Response
    FE-->>HR: Confirmation Message

🛠️ Technology Stack

Frontend

Technology	Version	Purpose
React	18.2.0	UI framework for building interactive interfaces
Material-UI	5.18.0	Pre-built components with Material Design
React Router	6.30.1	Client-side routing and navigation
Axios	1.12.2	HTTP client for API communication
React Simple Chatbot	0.6.1	Chat interface for Rasa integration

Backend

Technology	Version	Purpose
FastAPI	Latest	Modern, high-performance web framework
Python	3.11	Programming language
Supabase	Latest	PostgreSQL database with built-in authentication
Uvicorn	Latest	ASGI server for FastAPI
python-dotenv	Latest	Environment variable management

AI/ML

Technology	Version	Purpose
Sentence-Transformers	Latest	Semantic text similarity (all-mpnet-base-v2 model)
PyTorch	Latest	Deep learning backend
spaCy	Latest	NLP pipeline and named entity recognition
rapidfuzz	Latest	Fuzzy string matching for skill variants
LIME	Latest	Explainable AI (model interpretability)
Fairlearn	Latest	Bias detection and fairness metrics
scikit-learn	Latest	Traditional ML algorithms

Document Processing

Technology	Version	Purpose
pdfplumber	Latest	PDF text extraction
python-docx	Latest	Microsoft Word document parsing
pytesseract	Latest	OCR for image-based resumes
OpenCV	Latest	Image preprocessing for OCR

Chatbot

Technology	Version	Purpose
Rasa	3.x	Open-source conversational AI framework
Rasa SDK	Latest	Custom action server
TensorFlow	Latest	Deep learning backend for Rasa

Email & Notifications

Technology	Version	Purpose
smtplib	Built-in	Email sending via Gmail SMTP
python-jose	Latest	JWT token handling

📦 Installation

Prerequisites

Python 3.11 or higher
Node.js 18.x or higher
npm or yarn
Git

1. Clone the Repository

git clone https://github.com/Anvitha-git/AI-Driven-Resume-Screening.git
cd AI-Driven-Resume-Screening

2. Backend Setup

# Navigate to backend directory
cd backend

# Create virtual environment (Windows)
python -m venv venv
.\venv\Scripts\activate

# Create virtual environment (Linux/Mac)
python3 -m venv venv
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# Download spaCy model
python -m spacy download en_core_web_sm

# Create .env file
# Add the following variables:
# SUPABASE_URL=your_supabase_url
# SUPABASE_KEY=your_supabase_key
# JWT_SECRET=your_jwt_secret
# EMAIL_HOST=smtp.gmail.com
# EMAIL_PORT=587
# EMAIL_USE_TLS=True
# EMAIL_HOST_USER=your_email@gmail.com
# EMAIL_HOST_PASSWORD=your_gmail_app_password
# EMAIL_FROM_NAME=HR Team - AI Resume Screening System

# Run the server
uvicorn main:app --host 0.0.0.0 --port 8000 --reload

Backend API Documentation: http://localhost:8000/docs

3. Frontend Setup

# Navigate to frontend directory
cd frontend/c

# Install dependencies
npm install

# Start development server
npm start

Frontend Application: http://localhost:3000

4. Chatbot Setup

# Navigate to chatbot directory
cd chatbot

# Install Rasa (if not already installed)
pip install rasa

# Train the model
rasa train

# Terminal 1: Start action server
rasa run actions --port 5055

# Terminal 2: Start Rasa server
rasa run --enable-api --cors "*" --port 5005

Rasa API: http://localhost:5005

🚀 Usage

For HR Users

Register/Login
- Create an account with role "HR"
- Login with credentials
Create Job Posting
- Click "Create Job"
- Fill in job title, description, requirements, location, salary
- Submit to post the job
Upload Resumes
- Select a job posting
- Click "Upload Resumes"
- Choose multiple files (PDF, DOCX, PNG, JPEG)
- Wait for AI processing (~2-3 seconds per resume)
Review Rankings
- View candidates sorted by AI score (highest first)
- See color-coded scores:
  - 🟢 Green (80-100%): Excellent match
  - 🟡 Yellow (60-79%): Good match
  - 🔴 Red (0-59%): Poor match
- Click "Why this score?" to see explanation
Make Decisions
- Select decision from dropdown: Selected/Rejected/Pending
- Decisions saved automatically
- Click "Submit Decisions" to send emails to all candidates
View Explanation
- Click "Explain Score" button
- See breakdown:
  - Skill match contribution
  - Semantic similarity score
  - Experience evaluation
  - Education level
- View matched vs missing skills

For Candidates

Register/Login
- Create account with role "Candidate"
- Login with credentials
Track Applications
- View all job applications
- See status: Selected/Rejected/Pending
- Check ranking scores
Interview Preparation
- Click chatbot icon (bottom right)
- Practice common interview questions
- Get tips on:
  - STAR method for behavioral questions
  - Technical interview strategies
  - General interview advice
Receive Notifications
- Check notification bell for updates
- Receive emails when decisions are made

📷 Screenshots

Candidate Dashboard

Chatbot Conversation

HR Dashboard

Landing Page

LIME Explainability Example

Login Page

Resume Ranking View

Signup Page

📚 Documentation

Detailed documentation for each component is available in the docs/ folder:

Backend Documentation
- Technology stack and architecture
- AI/ML features (Sentence-BERT, LIME, Fairlearn)
- Resume parsing and ranking algorithm
- Email notification system
- API endpoints and database schema
- Why we chose these technologies
Frontend Documentation
- React component architecture
- State management with localStorage
- Material-UI integration
- Routing and navigation
- Decision workflow implementation
- UI/UX design patterns
Chatbot Documentation
- Rasa NLU pipeline and training data
- Conversation flow and intent classification
- Custom actions for interview prep
- Integration with frontend
- DIET classifier vs alternatives
Project Summary
- Quick reference guide
- Feature comparison
- Performance metrics
- Future roadmap

👥 Team

This project was developed as a Major Project at CMR Institute of Technology, AECS Layout, Bengaluru - 560037.

Team Members

Department of Artificial Intelligence and Machine Learning

Name	USN	Role	Responsibilities
Anvitha S	1CR22AI019	Frontend & Chatbot Developer	React Integration, Rasa Chatbot Implementation
Lingeshwari S	1CR22AI062	Frontend & Chatbot Developer	React UI, Rasa Chatbot UI
Pranav V	1CR22AI088	Backend & AI/ML Developer	FastAPI, AI Ranking Algorithm, Resume Processing
Shariq Sheikh	1CR22AI102	Backend & AI/ML Developer	Supabase Integration, Email Service, Bias Detection

Project Mentorship

Major Project Guide
Mr. Adarsh Benjamin
Associate Professor, Department of Artificial Intelligence and Machine Learning
CMR Institute of Technology

Head of Department
Dr. Shyam P Joy
Department of Artificial Intelligence and Machine Learning
CMR Institute of Technology

Academic Institution

CMR Institute of Technology
Department of Artificial Intelligence and Machine Learning
AECS Layout, Bengaluru - 560037
Karnataka, India

🎓 Project Highlights

AI/ML Innovations

Multi-factor Ranking Algorithm: Combines skill matching, semantic similarity, experience, and education
Explainable AI: LIME-based explanations for transparency
Bias Detection: Fairlearn integration ensures fair hiring practices
Semantic Search: Sentence-BERT (all-mpnet-base-v2) for context-aware matching

Technical Achievements

Full-stack Implementation: React + FastAPI + Rasa
Real-time Processing: < 3 seconds for 50 resume ranking
Scalable Architecture: Async FastAPI, client-side React routing
Production-ready: JWT auth, email notifications, error handling

User Experience

Intuitive UI: Material Design components
Responsive Design: Works on mobile, tablet, desktop
Interactive Chatbot: 24/7 interview preparation assistant
Transparent AI: Users understand why decisions were made

🔒 Security Features

✅ JWT token-based authentication
✅ Password hashing (Supabase)
✅ Role-based access control (HR/Candidate)
✅ TLS encryption for email (Gmail SMTP)
✅ Environment variable protection (.env)
✅ CORS configuration for API security

📊 Performance

Resume Parsing: < 2 seconds per file
AI Ranking: < 3 seconds for 50 resumes
Explanation Generation: < 1 second
Chatbot Response: < 300ms
Email Delivery: ~2 seconds per email

Interview scheduling integration (Google Calendar)
Mobile application (React Native)
Video interview analysis (facial expressions, tone)

Model Accuracy & Test Results

TABLE 1: OVERALL PROJECT ACCURACY

Metric	Value	Status
Overall Project Accuracy	87.1%	✅ Excellent
Expected Performance	70-80% (Industry Benchmark)	Above Average
Models Tested	4	Resume Screening, OCR, Fuzzy Matching, NER
Test Duration	5.45 seconds	Fast
Processing Speed per Resume	92.29 ms	Excellent

TABLE 2: COMPONENT-WISE ACCURACY (Weighted)

Component	Accuracy	Weight	Contribution	Status
Skill Extraction (NER)	93.62%	20%	18.7%	✅ Excellent
Fuzzy Skill Matching	87.50%	45%	39.4%	✅ Good
Semantic Similarity	80.00%	30%	24.0%	✅ Good
OCR (Image Processing)	100.00%	5%	5.0%	✅ Perfect
TOTAL WEIGHTED ACCURACY	87.1%	100%	87.1%	✅ Production Ready

Key Insights:

✅ Overall system accuracy of 87.1% exceeds industry benchmark (70-80%)
✅ Skill extraction performs excellently at 93.62% F1 score
✅ OCR achieves perfect 100% accuracy for image-based resumes
✅ Processing speed is 21x faster than target (92.29ms vs 2000ms)
✅ Production Ready - All components operating within acceptable ranges

Multi-language support for chatbot
Fine-tuned BERT model on resume datasets

📝 License

This project is licensed under the MIT License — see the LICENSE file for details.

🤝 Contributing

Contributions are welcome! This project is open for contributors — feel free to open issues or submit pull requests. Please follow the existing code style and include tests or documentation updates for non-trivial changes. For major changes, open an issue first to discuss the proposal.

🙏 Acknowledgments

Dr. Shyam P Joy, Head of Department, AI & ML, CMR Institute of Technology, for his guidance and support
Mr. Adarsh Benjamin, Associate Professor, AI & ML, CMR Institute of Technology, for his mentorship throughout the project
CMR Institute of Technology for providing resources and infrastructure
Open-source community for amazing libraries (React, FastAPI, Rasa)
Hugging Face for Sentence-Transformers models
Supabase for database and authentication services
Material-UI team for beautiful React components

📧 Contact

For queries or collaboration opportunities, please reach out:

Email: airesumescreening@gmail.com

Made with ❤️ by Team AI-Resume-Screening

CMR Institute of Technology, Bengaluru

Name		Name	Last commit message	Last commit date
Latest commit History 139 Commits
assets		assets
backend		backend
chatbot		chatbot
docs		docs
frontend		frontend
scripts		scripts
.gitignore		.gitignore
LICENSE		LICENSE
QUICK_START.md		QUICK_START.md
README.md		README.md
README_TESTING.md		README_TESTING.md
package-lock.json		package-lock.json
package.json		package.json

Folders and files

Latest commit

History

Repository files navigation

AI-Driven Resume Screening System with Chatbot Integration

📋 Table of Contents

🎯 Overview

Key Capabilities

✨ Features

For HR Professionals

For Candidates

🏗️ Architecture

Data Flow

🛠️ Technology Stack

Frontend

Backend

AI/ML

Document Processing

Chatbot

Email & Notifications

📦 Installation

Prerequisites

1. Clone the Repository

2. Backend Setup

3. Frontend Setup

4. Chatbot Setup

🚀 Usage

For HR Users

For Candidates

📷 Screenshots

📚 Documentation

👥 Team

Team Members

Project Mentorship

Academic Institution

🎓 Project Highlights

AI/ML Innovations

Technical Achievements

User Experience

🔒 Security Features

📊 Performance

Model Accuracy & Test Results

TABLE 1: OVERALL PROJECT ACCURACY

TABLE 2: COMPONENT-WISE ACCURACY (Weighted)

📝 License

🤝 Contributing

🙏 Acknowledgments

📧 Contact

About

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases 1

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Packages