Linkedin to Portfolio Site Generator

This project is a Python script that scrapes your Linkedin PDF and generates a customized portfolio site using OpenAI's GPT-4 model.

We interact with the GPT-4 model using LangChain.

1. Embed the Linkedin PDF
2. Store the embeddings into a Chroma vector database
3. Query that database to get relevant information
4. Generate text with OpenAI's GPT-4
5. With the generated text we use the Next JS portfolio site, powered by Nextra, to create the main portfolio file
6. Build and deploy site on Vercel

Demo

For this demo, I used LangChain's Co-founder and CEO Harrison Chase's LinkedIn.

You can find his deployed site here: https://harrison-six.vercel.app/

Video: https://www.youtube.com/watch?v=jY5UnSBq8sI

Prerequisites

Before you begin, ensure you have met the following requirements:

• Installed a recent version of Python (3.7 or newer) installed and a way to create virtual environments (virtualenv or conda)
• Created a Vercel account and have downloaded/login into Vercel CLI locally
• Created OpenAI API account and obtain an OpenAI API key

Getting Started

Clone the repo

git clone https://github.com/waseemhnyc/LinkedIn-to-Portfolio-Site-Generator

Create a virutalenv and source the environment

python3 -m venv myenv
source venv/bin/activate

Install the necessary libraries

pip install -r requirements.txt

Create a .env file and input your OpenAI API Key in the file

cp .env.example .env

Usage

To run the program, run the following command in the terminal:

python main.py

Ways to Improve

• Accept more inputs like resumes/CVs
• Grab data from Github and Twitter
• Integrate with other portfolio templates
• Push to Github so you could make your own changes
• Use LangChains output parser

Questions or Get in Touch

• Twitter: https://twitter.com/waseemhnyc
• Email: waseemh.nyc@gmail.com

License

This project is licensed under the MIT License - see the LICENSE file for details.