HireMind

HireMind is a NLP-powered resume analyzer that identifies how much job is aligned to resume. By extracting and comparing technical competencies, it then provides a "Match Score" and identifies exactly which skills the candidate is missing to pass the ATS.

Live at: https://hiremind-web.vercel.app/

FEATURES

• Resume PDF parsing using PyMuPDF
• Text preprocessing and normalization
• Skill extraction using predefined vocabulary
• Resume-job skill comparison
• Match Score calculation
• Identification of missing and matched skills

PIPELINE

The system follows a structured pipeline:

1. Resume Upload (PDF)
2. Text Extraction: Extract raw text from PDF
3. Text Cleaning: Normalize and preprocess text
4. Skill Extraction (Resume): Detect skills from resume
5. Skill Extraction (Job Description): Detect required skills
6. Skill Matching: Compare resume vs job skills
7. Scoring: Calculate match percentage
8. Output Generation: Display matched & missing skills

ARCHITECTURE

Backend follows a modular and layered architecture:

• api/ → Route handlers (FastAPI endpoints)
• services/ → Core business logic (Parser, Extractor, Matcher, Scorer)
• schemas/ → Data validation using Pydantic
• db/ → MongoDB integration and models

Pipeline-based design ensures separation of concerns and easier extensibility.

Frontend is built using Next.js with a component-driven structure and protected routes for authenticated access.

AUTHENTICATION FLOW

• User signs up or logs in via /auth
• Backend generates JWT token
• Token is stored on client side
• All protected routes require Authorization: Bearer <token>
• Backend validates token and extracts user identity for request handling

FILE HANDLING DESIGN

• Resume is uploaded as a PDF file
• Stored on server filesystem (/uploads)
• File path is stored in MongoDB
• Text is extracted using PyMuPDF and stored for fast reuse
• Enables separation between raw file storage and processed data

API DESIGN

• Authentication using JWT (issued on login, sent via Authorization header)
• Resume handled as PDF upload (not raw text input)
• Text is extracted and stored for analysis
• Analysis endpoint accepts job description as JSON payload

TECH STACK

• Backend: FastAPI (Python)
• Frontend: Next.js, Tailwind CSS
• NLP: Rule-based skill extraction (planned: embeddings for semantic matching)
• Auth: JWT-based authentication (HTTP Bearer tokens)
• Storage: Local file storage for resumes (PDF) with MongoDB metadata
• Deployment: Backend on Render, Frontend on Vercel

EXAMPLE OUTPUT

Match Score: 71.43%

Matched Skills:
- python
- docker
- aws
- sql
- kafka

Missing Skills:
- rest api
- microservices

Key Endpoints:

• POST /auth/signup → Register user
• POST /auth/login → Login & receive token
• POST /resume/ → Upload resume (PDF)
• GET /resume/ → View resume PDF
• GET /resume/text → View extracted resume text
• PUT /resume/ → Update resume
• DELETE /resume/ → Delete resume
• POST /analyze → Analyze resume vs job description

FUTURE IMPROVEMENTS

• Semantic skill matching using embeddings
• Improved skill extraction from project descriptions
• Resume rewriting suggestions
• Better UI/UX for visualization