🤖 AI Resume Matcher – 2025 Edition Capstone Project | Stevens Institute of Technology Team: Anugya Sharma, Maruthi Kunchala, Jash Shah, Yash Kalpesh Shah Duration: Jan 2025 – May 2025
📘 Overview The AI Resume Matcher is a full-stack intelligent platform that analyzes resumes and matches them to live job postings using natural language processing (NLP), machine learning, and real-time job feeds. It delivers personalized job recommendations, highlights skill gaps, and offers career advice through an integrated AI chatbot.
🔍 Features 📄 Resume Parsing & Skill Extraction
- Extracts text from .pdf and .docx resumes using PyPDF2 and docx2txt
- Uses regex and dictionary-based techniques to identify relevant skills
- Achieved 88% precision in skill extraction
🌐 Real-Time Job Fetching
- Integrated with JSearch, RemoteOK, and Remotive APIs
- Matches jobs dynamically based on parsed skill sets
🔗 Resume-to-Job Matching Logic
- Uses TF-IDF vectorization and cosine similarity to compare resume skills with job descriptions
- Ranks jobs and identifies missing skills for targeted improvement
📊 Job Clustering & Visualization
- Uses K-Means clustering and PCA for visualizing job market segments and alternate career paths
🤖 Career Advisor Chatbot
- Dual-model setup:
- FLAN-T5 for fast, private queries
- GPT-3.5 Turbo for deep contextual insights
- Provides contextual career advice based on resume and cluster data
📄 Downloadable Career Report
- Generates a clean, downloadable HTML report using Jinja2 templating
- Summarizes top job matches, skill gaps, and career recommendations
🛠 Tech Stack Languages & Frameworks: Python, Streamlit, Flask
- NLP & ML Libraries: spaCy, transformers, sentence-transformers, scikit-learn, torch
- Job APIs: JSearch, RemoteOK, Remotive
- Parsing Tools: PyPDF2, python-docx, regex, beautifulsoup4
- Visualization: Plotly, PCA
- AI Models: FLAN-T5, GPT-3.5 Turbo (via OpenAI API)
- Templating & Output: Jinja2, HTML
📦 Installation
-
Clone the repository: git clone https://github.com/Jash-stack/AI-Resume-Matcher-2025-Edition cd ai-resume-matcher
-
Create a virtual environment and install dependencies: pip install -r requirements.txt
-
Run the Streamlit app: streamlit run main_app.py
📈 Results
- Skill extraction achieved 88% precision
- Job relevance alignment was 76% compared to human reviewers
- Chatbot provided fast, context-aware responses across multiple use cases
🚧 Challenges & Future Work
- Improve handling of stylized/resume-formatted skills (e.g., ReactJS, Next.js)
- Incorporate feedback learning to improve ranking accuracy
- Enhance UI for better interactivity and accessibility
📬 Contact For questions or collaboration: 📧 jbs051814@gmail.com