T-Mobile Vibe Checker

Customer satisfaction and pending issues
Problem analysis
Sentiment analysis
Ratings compared with competition and outage map
Comparison with other carriers

Inspiration

Our team was inspired by the T-Mobile Challenge to go above and beyond, and offer advice and comparisons as well as data analysis

What it does

Our application allows T-Mobile to passively monitor customer sentiment in real time. We provide up-to-date recommendation on how to improve T-Mobile's service and how it fares against its competitors

How we built it

Our data analysis is built on the Nemotron model, which performs data analysis in a multi-agent setup, and acts as a chatbot using an Agentic RAG setup. We used a variety of scrapers for social media data, and stored the data in an Azure AI Search vector database. Our front-end is React-based with Vite elements, and is integrated with our backend using flask

Challenges we ran into

We had trouble with both the X API and Azure AI Search, due to the limitations of the free tier X API and free Azure search. We eventually had to drop X integration, but we managed to fit all the needed data partitions into the Free Azure Search

Accomplishments that we're proud of

We're proud of our integration of Azure AI Search with our Nemotron processing, as we have never done anything related to database management before

What we learned

We learned a lot from our first hackathon, but mainly we learned the skills needed to be a full-stack development team and how to balance practicality with imagination.