The app to create a cover letter based on the job description and your resume with LangChain + LLM This is the v0.0 version of the project.
foo@bar:~$ cd easycv
foo@bar:easycv$ pip install -e .Users should have an OpenAI key to have access to ChatGPT API. Please copy your resume to the "files" folder, and copy the job description to "files/jd.txt". Users can use the example prompt "user_cv_prompt.txt" or customize their own.
foo@bar:easycv$ python easycv.py --resume files/resume.pdf --jd files/jd.txt --user files/user_cv_prompt.txt -m gpt-3.5-turboThe prerequisites of installation are Python, pip, and Conda. If you find the installation failed because of the non-existent path of Conda, please install Miniconda or try to manually install those packages via pip.
If you see an error "Error Code 429 - You exceeded your current quota.", please check your plan and billing details. You need to increase your rate limit through OpenAI API page. Usually, 5 dollars is enough.
The tool uses OpenAI ChatGPT as the LLM to generate a cover letter.
The difficulty in this project is: beyond a single RAG system, how to find the matching information between the job description and the resume of the user.
I use a two-phase RAG QA to save the related contents in the chat history which is fed to the LLM along with user instructions to generate a cover letter.
- The Google Chrome extension user interface
- The REST API to call the backend "easycv" python script
- A few web scraper scripts for popular job searching websites
- Add an agent ask the user follow-up questions and use Google search tool