Chat_With_LinkedIn

📘 Overview:

For the project demo, visit this LinkedIn post.

This project aims to improve people's LinkedIn profiles and resumes using artificial intelligence. It will:

• 1️⃣ offer suggestions on how to strengthen your LinkedIn profile.
• 2️⃣ Help you create a professional resume tailored to your given job description.

---

📘 Project Goal:

To create an AI-powered tool that enhances LinkedIn profiles and generates tailored resumes based on user input and LinkedIn data.

---

📘 Technical Implementation:

• Foundation Model:

  The project was built on top of the Gemini-1.5-flash model API.

• Core Components:

  • LinkedIn-Content-Aware LLM: Interacts with users to provide LinkedIn profile improvement suggestions.
  • Resume-Aware LLM: Help creating resumes based on user-provided job descriptions.

• Data Processing:

  • Person's data is scraped from LinkedIn profile using Selenium, ChromeDriver, and BeautifulSoup.
  • Extracted information is structured into JSON format.

• Model Workflow:

  • The JSON data is used to prompt engineer the gemini-1.5-flash model for specific tasks used in the three directories (scrappers, linkedin_chatbot and resume_builder).
  • Langchain is employed to create LLM chains for conversation flow and resume generation.

• Deployment:

  • The prject is mainly run in GUI-mode (was designed using streamlit) on windows.
  • The final application was containerized by building a linux-based docker-image from DockerFile.

---

📘 How to use (You have 2 choices):

First, For Windows users:

• 1️⃣ run git clone https://github.com/OmarAllam22/Chat_With_LinkedIn.git
• 2️⃣ open the Chat_With_LinkedIn directory & run pip install requirements.txt
• 3️⃣ run streamlit run app.py

---

Second, For linux users via a docker image:

• 1️⃣ From your local machine's terminal, run xhost +.
  • This step makes your machine's X-server open to connections from any host (as we want to connect this server (located on your local machine) from inside the docker container).
  • This X-server is enabled for applications that needs to run GUI from inside docker container (as docker containers is mainly CLI-based).
• 2️⃣ From terminal, run docker build -t app_image .
• 3️⃣ From terminal, run docker run -it -v /tmp/.X11-unix:/tmp/.X11-unix:ro -e DISPLAY=$DISPLAY -p 80:8501 --name app app_image
  • This -v /tmp/.X11-unix:/tmp/.X11-unix:ro This mounts the local /tmp/.X11-unix directory to the container's /tmp/.X11-unix directory in read-only mode. This is typically used for running graphical applications within the container (GoogleChrome in our case).
  • This -p 80:8501 maps port 8501 inside the container (default port for streamlit) to port 80 on the local host machine (default port for localhost).
• 4️⃣ From your local machine, open a any web browser tab and search for http://localhost.

---

📘 Resources:

• Hands on LangChain tutorials. Here
• Docker By Ahmed Sami. Here
• Building GUI application with docker introducing X-server concept. Here
• Guide to building streamlit chatbot. Here
• About streamlit.session_state concept. Here
• Using chromedriver and selenium on linux. Here

---