InfraSight

Inspiration

We aimed to take a wide range of previous temperature data and past performance of towers to determine if there should be any attention given to any specific tower.

What it does

It displays towers in the downtown toronto area and uses sample data about the towers signal/frequency/and many other factors that are fed into a gpt API model to determine if a tower needs Maintenance or not.

How we built it

We first accumulated the data we would want to utilize, and a way to create more data to populate the database. We then created a custom prompt for the gpt model that is fed by at least 5 snapshots of datasets for any chosen tower. The towers are all displayed on a interactive map which when clicked displays various statistics about the tower as well as a the gpt prompt for recommended steps to take

Challenges we ran into

We had first wanted to use weather as a factor for prediction, but however all the cell towers we gathered were in the downtown Toronto area so the weather would be the same for all the towers. We also had trained a intuitive vertex ai model on a large sample dataset to predict if a cell tower is in good condition. However we wanted this project to focus more on genAI rather then ML/AI.

Accomplishments that we're proud of

We believe that this project idea is very inspired solves real-world problems that may exist for companies

What we learned

We have learned a lot about how to scale a project to the specifications required

What's next for InfraSight

InfraSight would like to expand to using weather data in its predictive models and by doing so expanding the cell towers we observe all across Canada and America.

Built With

• gpt
• javascript
• mongodb
• openai
• python
• react
• typescript
• vertex