resumejob / jobtracker Goto Github PK

View Code? Open in Web Editor NEW

4.0 2.0 3.0 133 KB

Organize your job application with python and chatbot

License: MIT License

Python 100.00%

jobtracker's Introduction

JobTracker

AI powered tool to trace job application status based on emails.

Feature

Easy to customize and use, you can adjust the confidence interval threshold yourself
Up to 96% accuracy with 0 false positives

Getting Started

Video Tutorial (Click to play at YouTube)

Prerequisites

Python 3.10
An OpenAI API key for ChatGPT access

Installation

Clone the repository and install the required dependencies:

git clone https://github.com/resumejob/JobTracker.git
cd JobTracker
# It is recommended to create a virtual environment for this project before install
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

Configure your OpenAI API key: Set up your API Key

Usage

Export your emails in MBOX format and run the script with the following command:

python3 main.py -p <path_to_your_email_file.mbox> -o <output_filename.csv>

Replace <path_to_your_email_file.mbox> with the path to your MBOX file and <output_filename.csv> with your desired output file name.

Example Output

You would get

checkout example_output CSV for more details

How It Works

Access Data
- Users export their emails in MBOX format from their email service provider.
Clean Up and Collect Data
- The system extracts basic information such as the sender, recipient, and body of the message.
- It uses predefined keywords from config.KEYWORD to determine if an email is related to a job application.
- If an email is related, the system collects the relevant information and forwards it to the ChatBot for analysis.
Understand and Format Data
- The ChatBot assesses whether the email pertains to a job application.
- For job-related emails, it identifies the current status of the application and suggests subsequent steps.
- A function call is utilized to neatly format the data.
Export Data
- The processed data is exported in CSV or Excel format for easy access and use by the job seeker.

Run Tests

Execute the following command to run tests:

python3 -m unittest discover -s tests

Roadmap

Distributing to PyPi
Support Local LLMs like Llama

Contributing

We welcome contributions from the community. If you would like to contribute, please fork the repository and submit a pull request.

Garen

Leveraged the ChatGPT/llama to process the email and a cost pipeline to estimate the price.
Merge email with same company and provide a clear view of the application timeline.
Increase model accuracy by adding a pre-process threashold.
Increase model speed by adding Hash to avoid the duplicate email.

License

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

Contact

For any queries, you can reach out to Project Maintainer.

jobtracker's People

Contributors

Stargazers

Watchers

Forkers

mive667 garenbbbb windsooon

jobtracker's Issues

Auto save the result in some iterations.

add estimated cost

If the user using ChatGPT, display the estimated cost and let the user confirm before processing

Just like interview-checklist, tell user a checklist when doing the next step. For example, the next step is "Send an email to abc@example to schedule a next-round HR call interview", the checklist may look like

Confirm interview details: Review the interview time, date, and platform to ensure you have the correct information and avoid any scheduling misunderstandings.
Review the job description: Understand the key requirements and responsibilities of the position to tailor your responses to the employer's needs.
Familiarize yourself with your resume and cover letter: Be prepared to discuss specific points or experiences in detail if asked by the interviewer.

You may use a fixed response text for different next step

Merge serveral same data into a single row

In example_output.csv, We can merge company and recipient_mail because they are the same

Compute the hash for each email to minimize redundant tasks

After using this tool, users may use it again after a period of time. At this point, emails that have been previously computed should not be recalculated. Users can specify a previously exported CSV file, and any emails that exist within it will not be recomputed.

Packaging and distributing to PyPi

Distribute the current project to PyPi so the user can install it using pip. Please have a look at Link

Add tutorial and example

text/video example to tell the user how to export mail to mbox format/import CSV to the notion or excel.
step by step video / text tutorial to walk through a basic whole example (with fake email content)

Improve classify accuracy

The prompt

"If this is a mail from a company I applied to or interviewed with before, use get_mail_info to get information"

It may incorrectly classify numerous emails that are unrelated to job applications. Try to improve accuracy by using another prompt or several prompts.(There are also other ways to improve it like to check if the sender ua se company domain in his/her email address

Add .env for API_KEY

Move API_KEY = 'your_api_key' environment variable from config.py to .env.

Use email date instead of date from Ai bot

We already get the date from utils.get_mail_info, we can use that date and put it in main.process_email instead of get date using ai bot

Support Llama

We would like to support Local LLMs like Llama, Have a look at

ollama which also supports API call
llama-cpp-python

Let the user use commend line args to choose LLM they want to use,

python3 main.py -l chatgpt

Merge state from the same company

In the current version, emails from the same company would be split into different rows, However, this is not what the user expects. It would be better to merge emails to the same row and only show the time in different states. For example

Company Name	State
ABC	- Applied: 7.12 - Next Round: 7.20 - Offer: 8.10

Add progress bar

Generate emails in MBOX format

We will add a script to generate different kinds of emails and store them in MBOX format. We can use this mbox file to test the accuracy of the tool.