AI Resume Scanner

Overview

This project is an AI-powered Resume Scanner that analyzes resumes and predicts job fields. It also recommends courses based on the predicted job category.

Implementation History

Initially, we implemented a TF-IDF + RandomForest Classifier approach for resume classification. However, the accuracy was not satisfactory, and we decided to implement BERT (Bidirectional Encoder Representations from Transformers) for better results.

1️⃣ TF-IDF + RandomForest Approach

• Used TF-IDF (Term Frequency - Inverse Document Frequency) for text vectorization.
• Trained a RandomForestClassifier on extracted resume text.
• While it provided a baseline model, its accuracy was limited due to TF-IDF's inability to capture contextual relationships.

2️⃣ BERT-Based Approach (Implementation Attempted)

• We implemented BERT for better contextual understanding of resume text.
• The model was loaded using transformers and torch, and fine-tuned on labeled resume data.
• However, due to dependency issues, OpenSSL problems, and conflicts with environment configurations, the BERT-based model did not run successfully.

Technologies Used

• Python (Primary programming language)
• Streamlit (For the web app UI)
• Scikit-learn (For TF-IDF vectorization and classification)
• Transformers & PyTorch (For BERT-based classification, but unresolved issues)
• PDFMiner (For extracting text from PDFs)
• SQLite (For storing user data and scores)
• Plotly (For visualization)
• Pafy & yt-dlp (For embedding YouTube career guidance videos)

Installation

1️⃣ Clone the repository

git clone https://github.com/your-repo/AI-Resume-Scanner.git
cd AI-Resume-Scanner

2️⃣ Create a Virtual Environment & Install Dependencies

python3 -m venv venv
source venv/bin/activate  # On Mac/Linux
venv\Scripts\activate  # On Windows
pip install -r requirements.txt

3️⃣ Run the App

streamlit run App.py

Next Steps

• Debug and resolve dependency issues to get BERT-based classification running.
• Improve resume scoring and recommendations.
• Optimize UI and performance.

Issues & Troubleshooting

Known Issues

• BERT Model Not Running: Encountered OpenSSL, numpy, and torch issues.
• pafy YouTube Integration: Requires yt-dlp instead of youtube-dl.
• Environment Conflicts: Dependencies need proper management with a fresh virtual environment.

Workarounds Tried

• Reinstalled openssl, numpy, and torch.
• Attempted using different pafy backends.
• Cleaned and rebuilt virtual environment.

Contributors

• Hadi Hijazi
• Community Contributions Welcome!

---

⚡ Let us know if you face issues or want to contribute! 🚀