ResuMix

Live Application: https://resu-mix.vercel.app/

A sentence-transformer-based AI-powered resume optimizer that generates job-specific, tailored resumes in seconds. Set your master resume once, then automatically generate perfectly optimized resumes for each job application.

What is ResuMix?

ResuMix is an intelligent resume optimization platform that uses advanced AI and natural language processing to help job seekers create highly targeted resumes. Instead of manually tailoring your resume for each job application, ResuMix:

Stores your master resume - Enter all your experiences, skills, and achievements once
Analyzes job descriptions - Uses sentence transformers to understand job requirements
Optimizes content - Intelligently selects and ranks your most relevant experiences
Generates PDFs - Creates professional LaTeX-formatted resumes ready to submit

Key Features

AI-Powered Optimization

Sentence Transformer Scoring: Uses the all-mpnet-base-v2 model to compute semantic similarity between your resume content and job requirements
Multi-Category Analysis: Evaluates content across technical skills, soft skills, and relevance
Smart Line Ranking: Each line in your resume is scored and ranked based on job fit

Master Resume Management

Structured Resume Builder: Organize your resume into sections, items, and lines
Multiple Item Styles: Support for various resume formats (2-4 heading styles, paragraphs)
LaTeX Export: Professional PDF generation using PyLaTeX
Parameter Customization: Fine-tune weights and biases for different content categories

Secure Authentication

Multiple Auth Methods: Email/password and Google OAuth integration
JWT-Based Sessions: Secure reauthentication tokens for quick access
PostgreSQL Storage: Secure user data and resume information storage

Modern User Interface

React + Vite Frontend: Fast, responsive single-page application
Intuitive Dashboard: Track resume status, tweak parameters, and generate optimized resumes
Real-time Feedback: Terminal-style output during resume generation
Wiki & Documentation: Built-in help system

Technical Architecture

Frontend

Framework: React 19 with Vite
Routing: React Router DOM
State Management: Context API for authentication
UI Components: Custom component library with Lucide icons
API Communication: Axios with credential support

Backend

Framework: Flask with Flask-RESTful
Language: Python 3.x
API Type: REST API
CORS: Configured for cross-origin requests from Vercel deployments

Database

Type: PostgreSQL
Connection: psycopg3 with connection pooling
Schema: Single data table storing user info and resume data as JSONB

CREATE TABLE data (
    uid VARCHAR(256) PRIMARY KEY,
    user_name VARCHAR(64),
    pwd VARCHAR(64),
    email VARCHAR(256) NOT NULL UNIQUE,
    auth_type VARCHAR(16),
    userinfo JSONB,
    resumeInfo JSONB
);

AI/ML Components

Sentence Transformers: all-mpnet-base-v2 for semantic similarity
OpenAI Integration: GPT-4o-mini for parsing job requirements and generating structured data
Vector Embeddings: Cached embeddings for efficient resume line scoring
Optimization Algorithm: Custom scoring and shuffling algorithm for content selection

Tech Stack

Backend Dependencies

Flask (3.1.1) - Web framework
sentence-transformers (4.1.0) - Semantic similarity
PyLaTeX (1.4.2) - PDF resume generation
PyJWT (2.10.1) - Authentication tokens
psycopg (3.2.3) - PostgreSQL database driver
openai (1.82.0) - AI integration
torch (2.7.1) - Deep learning framework
transformers (4.52.4) - NLP models

Frontend Dependencies

react (19.1.0) - UI framework
react-router-dom (6.30.1) - Routing
axios (1.10.0) - HTTP client
@react-oauth/google (0.12.2) - Google authentication
jwt-decode (4.0.0) - JWT handling
lucide-react (0.522.0) - Icons

How It Works

1. User Registration & Authentication

Users can sign up using email/password or Google OAuth. Authentication tokens are stored securely using JWT.

2. Master Resume Creation

Users build their comprehensive master resume by:

Adding sections (Education, Experience, Projects, Skills, etc.)
Creating items within sections (jobs, degrees, projects)
Writing lines for each item (bullet points, descriptions)
Setting parameters (category weights and biases)

3. Job Description Analysis

When generating an optimized resume:

User pastes a job description
GPT-4o-mini extracts 3-12 core requirements
Requirements are converted to vector embeddings
Each resume line is scored against requirements using cosine similarity

4. Optimization & Selection

The optimization algorithm:

Scores each line (0-10 scale) based on semantic similarity
Applies category weights and biases
Selects the most relevant content for each section
Generates multiple versions and shuffles for optimal layout

5. PDF Generation

Selected content is formatted using LaTeX templates
PyLaTeX generates a professional PDF
User downloads the tailored resume

Data Structure

Resume Hierarchy

ResumeInfo
├── sections (list)
│   ├── Section
│   │   ├── title (LaTeX string)
│   │   ├── items (list)
│   │   │   ├── Item
│   │   │   │   ├── titles (list of LaTeX strings)
│   │   │   │   ├── lines (list)
│   │   │   │   │   ├── Line
│   │   │   │   │   │   ├── content (LaTeX string)
│   │   │   │   │   │   ├── content_str (plain string)
│   │   │   │   │   │   ├── score (float 0-10)
│   │   │   │   │   │   ├── keywords (list)
│   │   │   │   │   │   └── aux_info (cached vectors)
│   │   │   │   ├── category_weight (dict)
│   │   │   │   └── category_bias (dict)
│   │   └── aux_info
└── heading_info (name, contact info)

API Endpoints

User Authentication

POST /user - Sign up new user
GET /user - Login with credentials or Google OAuth
PUT /user - Update user information

Resume Management

GET /resume - Retrieve user's resume data
POST /resume - Update resume information
PUT /resume - Modify resume parameters

Resume Generation

POST /resume/optimize - Generate optimized PDF resume
POST /api/generate-resume - Alternative PDF generation endpoint

Use Cases

Job Seekers: Generate tailored resumes for multiple job applications
Career Changers: Emphasize different skills for different industries
Recent Graduates: Optimize limited experience for various roles
Professionals: Maintain one master resume, generate role-specific versions

Security Features

Password hashing for email/password authentication
JWT tokens with expiration for session management
Environment variable management for API keys
CORS configuration for secure cross-origin requests
PostgreSQL with parameterized queries to prevent SQL injection

Future Enhancements

Multiple resume templates and styles
ATS (Applicant Tracking System) optimization
Cover letter generation
Interview preparation based on job requirements
Analytics on which experiences perform best
Multi-language support

Contributing

This project is actively maintained. For questions or contributions, please refer to the project repository.

License

See repository for license information.

Acknowledgments

Sentence Transformers library by UKPLab
OpenAI GPT models
PyLaTeX for PDF generation
The open-source community

Built to help you land your dream job faster.

RULE:

usually, make means to parse the information and to get the thing ready to be built
build would be to build the object.

UPDATE: scoring is done only once with the master scoring function to enable functions such as:

balancing skills
preference of variety or depth

Name		Name	Last commit message	Last commit date
backend		backend
frontend		frontend
.gitattributes		.gitattributes
.gitignore		.gitignore
README.md		README.md
REAUTH_IMPLEMENTATION.md		REAUTH_IMPLEMENTATION.md
TODO.txt		TODO.txt
debug_plan.txt		debug_plan.txt
grass		grass
package-lock.json		package-lock.json
package.json		package.json
requirements.txt		requirements.txt
resume_test_output.pdf		resume_test_output.pdf
revamp_handoff.txt		revamp_handoff.txt
test_api.sh		test_api.sh
test_resume.pdf		test_resume.pdf
test_resume2.pdf		test_resume2.pdf
test_seg.sh		test_seg.sh
vercel.json		vercel.json

Latest commit

History

Folders and files

Repository files navigation