J.O.B.S — Job Opportunity Bot System

---

🛠️ Overview

J.O.B.S is a system designed to automate the process of:

• Fetching and parsing job-related emails
• Cleaning email bodies for consistent processing
• Classifying which emails are relevant to job applications
• Extracting structured job information (company name, role, application status)

The goal is to make job application tracking more reliable and less manual.

---

Completed Tasks

• Built base project framework
• Created configuration system
• Implemented EmailHandler class for fetching and parsing emails
• Developed email body cleaner for classification prep
• Created agent for classifying job-related emails
• Created agent for extracting job opportunity information
• Created a much better cleaning logic
• Created an agent for further cleaning of data. Can removed if using a larger model but 3.5-turbo tends to misclassify without this
• Implement spreadsheet insertion function

---

Upcoming Tasks

---

Future Improvements (optional)

• Auto-detect and handle ghosted applications
• Track application stages (applied, interview, offer, rejected)
• Expand support for more job platforms (Indeed, Glassdoor, Wellfound)

---

⚡ Project Notes

• Focus: Start with LinkedIn emails first (most common and most inconsistent)
• Gmail API Mode: Testing format=full instead of raw to avoid unnecessary manual decoding
• Parser Design: Sender-based cleaning to handle differences across platforms
• Goal: Keep processing efficient, avoid wasting tokens or resources unnecessarily

---

Known Challenges

• LinkedIn emails are inconsistently formatted and contain a lot of tracking artifacts
• Some recruiters use customized templates that may require special handling
• Email bodies often contain embedded noise even after initial decoding

---

Current Phase:

LinkedIn-only cleanup and parsing focus, building a scalable structure to later support multiple job sources. Creation and updation of job spreadsheet on gdrive