Build your resume once ONLY!!

"Enhance" your experience to increase your chances of landing interviews, while saving valuable time for side projects and learning.

Auto Resume

This repository seeks to harness the capabilities of ChatGPT in generating more captivating and job description-targeted resumes from your base resume. I realized that the conversion rate from sending resumes to securing interviews was exceedingly low (around 2-3% per 100 applications). Enhancing your chances of an interview requires tailoring your resume to match the job description (e.g., incorporating relevant keywords like 'Python' for a python developer role). I found this "resume-fitting" task to be extremely annoying and meaningless. As a software developer, you'd likely prefer to devote your time to creating your next impactful project to address real-world issues—utilizing your software engineering skills to improve the lives of others.

Consequently, I am motivated to develop a solution that amplifies the likelihood of landing an interview while preserving every developer's valuable time.

Current state

I found that the auto mode did not make a good job. Most of the time the generated bullet points did not make sense. Therefore for the moment I created a dash app to version control your resume. I simply use the ChatGPT web interface to generate bullet points and I only stored those good generation in my database. You can use add experience on the app to do so.

Interface mode

Setup

Use poetry to setup the environment

poetry install

Use dash app to version control your resume and experience generateion

Create a .env file with reference to .env.sample
Prepare your base resume json
Seed your sqlite DB

# 
python seed.py

Use CLI to generate your resume automatically

Set the Open AI api key as environment variable

# add this line in your .zshrc / .bashrc
export OPENAI_API_KEY="your-openai-api-key"

Prepare your base resume json Prepare a base resume in JSON. Please refer to the resume.example.json
Prepare the JD of your targeted job opening Prepare a txt file containing the JD of your targeted job opening.
Run with the CLI

python generate_resume.py --resume your-base-resume-json.json --jd your-job-description.txt

Documentation

Last update: 2023-11-17

Agents

1. jd_parsing_agent

This agent parses a given job description to extract key information: the job title, skills required, and other important keywords. The agent will use job descriptions as input and format the extracted data into a dict.

job_description: str, the actual job description.

2. keyword_injecting_agent

This agent enhances a set of job experiences by incorporating specific keywords derived from a job description. The goal is to make the experiences more tailored to the job description provided. The agent will review user's experiences, selecting relevant keywords, and then rewriting those experiences with the chosen keywords.

experiences: str, the user's job experiences. keywords: list, the list of keywords extracted from a job description

3. skill_injecting_agent

This agent revises a set of job experiences by including certain required skills. The agent aims to tailor the experiences to better fit the job requirements by including relevant skills. The agent will review the experiences, select the most relevant skills, and rewrite the experiences with these skills included.

experiences: str, the user's job experiences. skills: str, the list of required skills.

4. experience_enhancing_agent

This agent enhance job experiences provided from a user's job history. The objective is to refine these experiences according to specified criteria to potentially better reflect the user's accomplishments and skills. The agent ensures the experiences are action-oriented, job duty clear, skill-inclusive, and achievement-focused with quantifiable outcomes.

experiences: str, the user's job experiences.

Interaction logic of agents

The engine_v1.py employs a multi-agent system to refine a user's resume according to a given job description. The start() method encapsulates the logic of building the resume and the create_agents() method initializes all agents. Here's how the agents interact and contribute to the resume enhancement process:

1. Initialization:

All agents are initialized with the same LLM (Language Model) configuration.

2. Job Description Parsing:

The JdParsingAgent reads the job description and extracts essential details like required skills and keywords.

3. Skill Injection:

For each job experience in the base resume, the SkillInjectingAgent incorporates relevant skills from the job description into the job experience description.

4. Keyword Injection:

Subsequently, the KeywordInjectingAgent incorporates relevant keywords from the job description into the enhanced job experience description.

5. Experience Enhancement:

Lastly, the ExperienceEnhancingAgent further refines the job experience descriptions by ensuring they are action-oriented, clear, skill-inclusive, and achievement-focused with quantifiable outcomes.

6. Resume Update and Output:

The updated job experiences are collated and used to update the original base resume. A utility function create_resume() is then employed to format the updated resume into a docx file, which is saved to the specified output file path.

References

Canadian resume template; https://zety.com/blog/canadian-resume

Feedback and Contributions

Your feedback on the quality and effectiveness of this solution is highly appreciated. I am eager to learn and improve from your insights! Additionally, feel free to make pull requests—let's collaborate to enhance the hiring process together!

lokhiufung/auto-resume