AVIRI - Authentic Virtual Identity Recruitment Interface

AVIRI transforms hiring by converting resumes into engaging AI-driven video pitches. Recruiters interact with candidate avatars via real-time chat, making talent discovery fast, visual, and interactive.

🔗 Video Demo: Google Drive

---

🔍 Problem

Recruiters spend too much time on manual resume review, repetitive calls, and outdated systems. AVIRI streamlines this with a smart, visual, and interactive hiring experience that feels as easy as scrolling reels.

---

🚀 What It Does

• Generates video avatars from resumes and photos
• Provides a chat interface powered by Hugging Face for recruiter-agent interaction
• Displays candidates in a swipe-style carousel UI
• Supports bookmarking, liking, messaging, and dark mode
• Enables inclusive hiring with accessibility and multilingual pitch support

---

💡 Key Features

• Resume Parsing: Converts resume into structured profile info
• Pitch Video Generation: Uses SadTalker, EdgeTTS, FFmpeg for video synthesis
• Avatar Chat: Hugging Face models power real-time AI conversations
• Carousel UI: Swipe left/right to shortlist or reject candidates
• Background Removal: rembg and face_recognition for clean avatar videos
• Inclusive UI: Accessibility mode, dark mode, multilingual agents

---

🛠️ Tech Stack

Component	Tech Used
Frontend	React.js
Backend	Node.js, Express.js (Nodemon)
Database	MongoDB
Resume Parsing	Python
Chatbot Integration	Hugging Face Transformers (LLM models)
Video Generation	SadTalker, EdgeTTS, FFmpeg
Image Processing	rembg, face_recognition

---

📁 Folder Structure

LinkedIn-Hack/

├── Avatar/ # Talking avatar

generation (SadTalker, etc.)

├── backend/ # Node.js backend

│ ├── app.js

│ └── routes/

│ └── controllers/

├── frontend/ # React-based

frontend

├── models/

│ └── Elevator pitch/ # Parsed

resume text , pitch scripts

---

🧪 How to Run

1. Clone the repository

git clone https://github.com/itsvamz/LinkedIn-Hack.git
cd LinkedIn-Hack

2. Backend setup

cd backend
npm install
npx nodemon app.js

2. Frontend setup

cd frontend
npm install
npm run dev

⚠️ Make sure to configure .env with MongoDB URI, Hugging Face access token, and other credentials.

📦 Key Dependencies

react, axios, tailwindcss

express, mongoose, nodemon

huggingface, transformers, python-shell

formidable, ffmpeg-static, sadtalker

edgetts, rembg, face_recognition

✅ Future Enhancements Real-time live agent interviews

QR code to launch pitch on mobile

Dynamic pitch updates over time

Blockchain-based credential verification

---

LinkedIn-Hack: Full Stack App Deployment & Usage Guide

1. Environment Setup

A. Setting up .env

• Copy .env.example to .env in both backend and frontend folders.
• Fill in all required values (DB URIs, API keys, service URLs, etc.).

B. Downloading Model Files

• Install gdown (if not already):

  pip install gdown
  

• Run the model download script from the project root:

  bash scripts/download_models.sh
  

• This will download all required models to Avatar/checkpoints/.

2. Running the App Locally

A. Backend

• Install dependencies:

  cd backend
  npm install
  

• Start the backend:

  npm start
  

B. Avatar (SadTalker) Service

• Install dependencies:

  cd Avatar
  # (Optional) Create and activate a virtual environment
  # python3 -m venv venv
  # source venv/bin/activate  (Linux/Mac)
  # venv\Scripts\activate    (Windows)
  pip install -r requirements.txt
  

• (If needed) Start the Gradio demo UI:

  python app_sadtalker.py
  

• For backend integration, no need to run a separate service; the backend calls the Python script directly.

C. Frontend

• Install dependencies:

  cd frontend
  npm install
  

• Start the frontend:

  npm start
  

3. Platform-Specific Notes

• Windows: Use venv\Scripts\activate to activate Python virtual environments.
• Linux/Mac: Use source venv/bin/activate.
• Always run the model download script from the project root to ensure files go to the correct directory.

4. Deployment Plan

A. Choose Hosting Platforms

• Frontend: Vercel, Netlify, or similar.
• Backend: Render.com, Railway, Heroku, or a cloud VM (AWS, Azure, GCP).
• Avatar (Python):
  • If backend and avatar are on the same server, no extra step.
  • For scaling, deploy avatar as a separate service (Render, Railway, or a VM).

B. Prepare for Deployment

• Frontend:
  • Build with:

    npm run build
    

  • Deploy the build/ folder to your chosen platform.
  • Set environment variables (API URLs) in the platform dashboard.
• Backend:
  • Zip and upload your backend folder (excluding node_modules and large files).
  • Set all environment variables in the platform dashboard.
  • Run the model download script on the server after deployment.
• Avatar:
  • If needed, run the model download script and install dependencies on the server.

C. Media Storage

• For production, use a cloud storage provider (AWS S3, Cloudinary, ImageKit, etc.) for uploads.
• Update your backend to use cloud storage for avatars, resumes, and videos.

5. Testing Checklist

After deployment, test the following:

• Registration and login
• Resume parsing
• Avatar/video/photo upload and generation
• All user flows (end-to-end)

---

For any issues, check logs on your deployment platform and ensure all environment variables and model files are correctly set up.

AvatarAI