Payment Page
Cafe Verona
Main page

Inspiration

Last month, part of our group was in Japan. While there, we had a difficult time finding free wifi networks that we could hop onto and had to ask whenever we went into a business if they had wifi. Using this as inspiration, we decided to create Roam.

What it does

Roam is an application where businesses can choose to add their wifi and charge consumers a small fee for a preset timeslot. Customers can then access a Google Maps overview of the companies in their area that are opted into the application.

How we built it

We used a Raspberry Pi 5 to route our network traffic through. We also created a react native mobile app for IOS / Android users. On the app, we use the Google Maps API to display icons of businesses that offer the wifi for purchase service. Users can click an icon and submit a payment through the Stripe API to access the wifi network for a limited time if they are within range.

Challenges we ran into

Getting the Pi working -DHCP config -There were a ton of challenges in getting an SLM working. Both due to the hardware limitation and the software challenges. One of our members was able to get the qwen2.5-1.5b-instruct-q4_k_m.gguf model running locally, but there's a ton of config work that needs to change when using the Pi. We really wanted to have the model ingest network information that we capture on the Pi and make management decisions based on it- identify bad actors, throttle traffic, boot from the network, give traffic analytics, and allow the business owners to type (and eventually use voice) plain text commands and have the model generate a script that would accomplish that task. For example-- stop traffic from TikTok, remove this device, show me my network utilization rate, etc.

Accomplishments that we're proud of

Network tunnel through the Pi so we can monitor and control traffic and have devices connect -Stripe integration -SLM up and running locally, even if not fully integrated into the final project with system commands

What we learned

Don't bite off more than you can chew. We were really ambitious and got pretty far. Still, we basically tried cramming 2 projects into 24 hours- the actual business application side of the project and the SLM network manager side of the project.

What's next for Roam

-Optimize SLM to run on the Pi and ingest and USE network information to help optimize and manage the network -Voice integration -Stream network insights to business devices -Web app