1. The Challenge (Problem Solved) T-Mobile requires a solution to move from reactive, slow reporting to proactive, real-time intelligence. The challenge is connecting two disparate data sources—customer emotion (sentiment) and operational reality (network health),to instantly flag issues and prevent major escalations.

  2. The Solution (HappyConnect) I built HappyConnect: The T-Mobile Customer Happiness Index, a cloud-ready platform that streams data to a unified, real-time dashboard. The core value is the Gemini AI Executive Summary.

When a user identifies a problem period on the chart (e.g., a spike in negative sentiment), they click the 'Analyze' button, and the Gemini API instantly:

Analyzes the filtered data (sentiment scores, time, network metrics).

Provides a prioritized, single-paragraph action plan that identifies the root cause and recommends immediate operational steps.

This instantly turns a visualization tool into a decision-support system.

  1. Key Technical Features

Multi-Threaded Architecture: The data pipeline runs on a separate, asynchronous thread (data_pipeline.py). This is a critical engineering feature that ensures the web server (Dash/Flask) is always responsive and never freezes, even while processing continuous data updates.

Modular & Cloud-Ready Codebase: The application is organized into five clean, modular Python files (app.py, ai_client.py, etc.), making it easy to test, maintain, and deploy to the cloud (via Docker/containers).

Contextual Gemini Integration: The ai_client.py module is engineered to send the specific, filtered data context to the Gemini model, resulting in hyper-relevant, actionable analysis instead of generic summaries.

  1. Gemini AI Impact

Gemini transforms the operational workflow. It eliminates the hours of manual analysis previously needed to diagnose an issue. By providing an instant executive action plan, HappyConnect empowers T-Mobile teams to enhance the CX by responding to incidents in minutes, not days.

Built With

Share this project:

Updates