applycrew

A multi-agent pipeline that drafts a tailored job application response from a job description and your resume.

What it does

Given a job description and a resume (both as plain text files), applycrew runs them through five LLM agents in sequence and outputs a copy-paste-ready application response:

1. JD agent — extracts key skills, required experience, technologies, and soft skills from the job description.
2. Resume agent — extracts strongest accomplishments, relevant technical skills, experience, and quantified impact from the resume.
3. Fit agent — compares the two analyses to find matches, gaps, the best positioning angle, and proof points to emphasize.
4. Writer agent — drafts a concise, natural application response based on the fit analysis.
5. Reviewer agent — polishes the draft using the original JD and resume as context, removing buzzwords and AI-sounding phrasing, and returns the final text.

How to use it

1. Clone and set up a virtual environment

git clone https://github.com/dachi-dev/applycrew.git
cd applycrew
python3 -m venv venv
source venv/bin/activate
pip install langgraph langchain-openai langchain-core python-dotenv

2. Configure environment variables

Create a .env file in the project root:

LLM_MODEL=llama-3.3-70b-versatile
LLM_BASE_URL=https://api.groq.com/openai/v1
GROQ_API_KEY=your_api_key_here

The pipeline uses ChatOpenAI pointed at any OpenAI-compatible endpoint. Defaults above use Groq (free tier). To use OpenAI, Together, OpenRouter, or another provider, change LLM_MODEL and LLM_BASE_URL and use the matching API key.

3. Add your job description and resume

Replace the templates in:

• data/job_description.txt — paste the full job description
• data/resume.txt — paste your resume

4. Run

python3 main.py

The final application response is printed to stdout.

Technologies

• Python 3
• LangGraph — orchestrates the agent pipeline as a directed state graph.
• LangChain (langchain-openai, langchain-core) — prompt templates and LLM client.
• python-dotenv — loads environment variables from .env.
• OpenAI-compatible LLM backend — defaults to Groq's Llama 3.3 70B, but works with any OpenAI-compatible API.

Project structure

applycrew/
├── agents/
│   ├── jd_agent.py
│   ├── resume_agent.py
│   ├── fit_agent.py
│   ├── writer_agent.py
│   └── reviewer_agent.py
├── data/
│   ├── job_description.txt
│   └── resume.txt
├── graph.py        # LangGraph pipeline definition
├── main.py         # Entry point
└── .env            # Not committed