maulana
All projects

WhatsApp leave sync

Converts staff leave and attendance reports received over WhatsApp into attendance records. Uses a Node.js WhatsApp listener, Gemini-based extraction, and Python reconciliation with an existing Microsoft Access database, including duplicate and revoked-message handling.

Internal

Internal tool

Automation / AI/ML

Internal project — no public repository or demo.

Leave synchronization
WhatsAppGemini + validationAttendance records
Workflow illustration / Internal tool
Architecture illustration based on the supplied implementation. No live operational data is shown.

Problem

Staff report leave and absences over a WhatsApp group; someone then had to read every message and update the attendance database by hand.

Approach

A Node.js listener (Baileys) watches the group and forwards messages to Gemini, which extracts structured leave records (validated with Pydantic). A Python reconciliation step writes them into the Microsoft Access attendance database via PyODBC — matching staff, ignoring duplicates, and deleting records when a WhatsApp message is revoked. PM2 keeps both sides running.

Implemented features

  • WhatsApp message capture via a Node.js Baileys listener
  • Gemini-based extraction of unstructured leave messages
  • Pydantic-validated records written to Microsoft Access
  • Duplicate detection and revoked-message cleanup

Availability

Internal project — no public repository or demo.