Russell Day - Education Developer 2024-2025

MDST_Email_Manager_Demo (2).mp4

Background

In managing our club's growing activities, we quickly became overwhelmed by the high volume of incoming emails. Individually, each email was relatively simple to address, but collectively, they resulted in significant backlog and delays in our responses, reducing our overall efficiency. It was clear we needed a more effective system. So Weile, seeing my skillset actually reached out to me to start and run a project to figure out a possible solution to our issue.

My Innovative Solution

So, how did I approach this task? Initially, I considered a simple manual process—copying emails into ChatGPT for responses. Although straightforward, I realized this would quickly become tedious. But since I enjoy tackling these sorts of challenges, I decided to develop a more robust and automated solution. Here's what I envisioned:

• Email Fetching: Emails are automatically fetched from Gmail, leveraging Google's OAuth authentication to ensure secure access.
• Automated Response Generation: Fetched emails are sent to Groq's Inference API, powered by the advanced LLAMA 3.3 model, generating context-aware, personalized responses.
• Human Review: Generated responses are saved and presented in an easy-to-navigate interface, allowing club members and myself to review, edit, and approve responses before manually sending them.

This workflow streamlines our communication without losing the personal touch essential to our interactions.

Frontend Interface

To keep the solution user-friendly, I built the frontend using Streamlit, featuring three key components:

• MDST Manager: A central dashboard where users can fetch new emails, generate automated responses, and manage drafts easily.