Local ATS (Applicant Tracking System) with Python & FAISS

This project is a local Applicant Tracking System that:

• Parses multiple PDF resumes.
• Lets you paste a job description.
• Scores resumes against the job description using semantic similarity.
• Runs completely offline with a local embedding model (model.safetensors).

---

Features

• Offline: No cloud AI calls — works fully on your machine.
• Fast: Uses all-MiniLM-L6-v2 or any local sentence-transformers model.
• Scalable: Can search thousands of resumes using FAISS.
• Easy to Use: Just drop your resumes in a folder and run.

---

Requirements

• Python 3.8 or higher
• A local embedding model in model/ (must include: model.safetensors, config.json, tokenizer.json, tokenizer_config.json)

---

Installation

1. Clone the repository

git clone https://github.com/Pepperjack-svg/ATS.git
cd ATS

2. Create a virtual environment

python -m venv venv
# Activate:
# Windows:
venv\Scripts\activate
# macOS/Linux:
source venv/bin/activate

3. Install dependencies

pip install --upgrade pip
pip install pdfplumber sentence-transformers faiss-cpu

---

Usage

1. Place resumes in the resumes/ folder.
2. Make sure your local embedding model is in the model/ folder.
3. Run:

python app.py

4. Paste your job description when prompted (press Enter twice to finish).
5. Get a ranked list of matching resumes.

---

Example Output

[INFO] Loading embedding model from model ...
Paste your job description below (press Enter twice to finish):
Looking for a Python developer with experience in Django and REST APIs.

[INFO] Reading resumes...
[INFO] Encoding resumes...
[INFO] Scoring resumes...

=== ATS Match Results ===
1. resume.pdf — Match Score: 89.42%

---

Notes

• If you only have model.safetensors, download the rest of the model files from Hugging Face.
• To keep your repo small, add venv/ and model/ to .gitignore.

---