CV_ALIGN is an intelligent, full-stack platform that revolutionizes the way companies evaluate resumes. It enables recruiters to post jobs and candidates to apply by uploading CVs. These resumes are automatically scored and analyzed using a hybrid AI approach combining semantic similarity, keyword mapping, and LLM-based feedback.
Companies can then view ranked applications with detailed feedback and make informed accept/reject decisions — which are instantly reflected on the candidate's dashboard. This eliminates the hassle of manual screening and makes the recruitment process seamless.
- 🔐 Authentication System for both recruiters and candidates
- 🧾 Job Posting & Resume Submission with Cloudinary-powered upload
- 📊 Automated Resume Evaluation using AI models
- 📈 Score-based Ranking of candidates per job
- 💬 LLM-generated Feedback on resume strengths and weaknesses (via Google Gemini)
- ✅ Recruiters can accept/reject applications from their dashboard
- 🔄 Candidates get real-time status updates on their submissions
- React + Vite – lightweight and fast SPA setup
- Vanilla CSS – for custom UI styling
- React Router DOM – for routing
- FastAPI – modern Python web framework
- Python 3.x
- SQLite (can be swapped with PostgreSQL/MySQL)
- SQLAlchemy – ORM for schema and queries
- JWT (JSON Web Tokens) – for secure login sessions
- bcrypt – for password hashing
- Cloudinary – for uploading resumes and job descriptions
- Sentence Transformers – semantic similarity scoring
- TF-IDF Matching – keyword-based relevance scoring
- LangChain + Google Gemini – LLM-based feedback generation
- spaCy, Regex, NER – for information extraction
- PyPDF2 / Mammoth / Textract – for file parsing
CV_ALIGN/
├── Backend/
│ └── app/ # FastAPI backend app
│
├── Frontend/
│ └── my-app/ # Vite + React frontend
│
├── Model/
│ └── new_main.py # Resume evaluation & scoring logic
│
├── requirements.txt # Python dependencies
└── README.md
Resume File
│
├─▶ 1. Document Ingestion & OCR Parsing
│ (PyPDF2, Mammoth, Textract, Tesseract)
│
├─▶ 2. Information Extraction
│ (Regex, spaCy, NER)
│
├─▶ 3. Preprocessing & Synonym Mapping
│ (custom JSON maps + keyword expansion)
│
├─▶ 4. TF-IDF + Embedding Similarity
│ (Sentence Transformers + cosine similarity)
│
├─▶ 5. Experience Scoring
│ (custom logic on titles & dates)
│
├─▶ 6. LLM Verdict via LangChain + Gemini
cd CV_ALIGN/Backend
python -m venv venv
source venv/bin/activate # Windows: venv\Scripts\activate
pip install -r requirements.txt
uvicorn app.main:app --reloadServer runs at: http://127.0.0.1:8000
cd CV_ALIGN/Frontend/my-app
npm install
npm run devApp runs at: http://localhost:5173
Model is auto-triggered on /apply, but for testing:
cd CV_ALIGN/Model
pip install -r requirements.txt
python new_main.pyThis project is currently not licensed.