RowdyPay

Onboarding
Your profile
Your groups
Group details
Adding a budget request
Budget requests can use OCR
Confirming the request
Review your reliability report

Inspiration

Have you ever gone out to eat with the same friends only to have to coordinate how to split the bill properly for the seven hundredth time? It's such a pain. Who pays who? How much do I owe you? How should I pay you? Therefore, we created RowdyPay, a mobile budgeting solution that seeks to end this inconvenience once and for all.

What it does

RowdyPay is an iOS app that allows groups of users to manage shared budgets. Users can scan receipts and utilize Optical Character Recognition to parse the details. From there, users can automatically send a payment request to a shared budget, where other users in the group can then contribute their share.

How we built it

RowdyPay uses SwiftUI for a mobile iOS frontend, which connects via a REST API to a blazingly fast backend written entirely in Rust. We used one of our team's laptops as a server and took advantage of the OpenAI API to perform OCR on pictures of receipts.

Challenges we ran into

Early on, we knew it was imperative that we had the proper setup for the rest of the project. We had to spend meticulous effort in setting up the middleware correctly the first time around, which meant there were more than a few refactorings.

Accomplishments that we're proud of

We are incredibly proud of the functionality of our app. It's a working product, done right! We have a beautiful front end paired with a robust and multithreaded backend, with no loose screws left anywhere.

What we learned

We learned that communication is absolutely essential, especially when working in projects that have many moving parts. Several times we had to pause our work and orient ourselves in order to keep moving in the right direction.

What's next for RowdyPay

We hope that we can polish RowdyPay into a mobile app worthy of the Apple App Store!

Built With

ocr
openai
rest-api
rust
swiftui

Submitted to

RowdyHacks X
- Winner Overall Track: 1st Place

Created by

I was one of the two front end developers on the project. My primary responsibility was to build the middleware for the project which communicated between the front and back ends. This meant building all underlying object structures in Swift and writing functions to convert them from JSON received from the various REST API requests made to the server. I was also tasked with using the OpenAI API to read receipts using Optical Character Recognition, which was difficult to create the request and then retrieve the result in a meaningful way. And, of course, I also had a hand in the design and structure of many of the app's views. Overall, it was a tedious process to get everything right, but it (hopefully) all works now.

Aiden Seibel
I worked on the UI design and user experience enhancement, focusing on creating smooth navigation flows and implementing interactive animations throughout the app. I also integrated functionalities relating to group and payment systems through consuming data from Sam's backend endpoints ensuring a seamless connection between our server and UI. My implementation emphasized on clean error handling, responsive loading states, and consistent layouts across all views. I payed deep attention on organizing designs and navigation in a way that makes the payment process intuitive for users. Working alongside Sam's Rust backend and Aiden's middleware, my frontend contributions helped create a balanced full-stack application that combines technical capability with user-friendly design.

khoitran999
I worked as the backend developer on this project and decided to explore a different direction than something like a Node.js backend; Axum with Rust. Rust has been my favorite language for as long as I have been learning it and was super excited when I was presented with a perfect opportunity to utilize this unique tool. I decided on the Axum framework because of it's lightweight and low abstraction methods which allow for basic http connection all the way to postgres database connection and persistent data manipulation. Rust's strong typing proved to be incredibly beneficial although at times challenging. Dealing with DECIMAL types when inserting data into tables lead me down the path of exploring the sqlx BigDecimal type and learning more about type casting and conversion between Rust Primitive types and more abstract types. Another challenge I faced was managing what I had hoped to be a very modular project with a Rust workspace being the heart of the design. This design consisted of an api binary crate and library crates for commands, controllers, data transfer objects, models, and utils. While this initially went smoothly, I ran into an issue just about halfway through the hackathon when attempting to connect and share my postgresql database pool with my controller logic. It turns out that there is very limited support for sharing state from an Axum router across Rust workspaces; this is something that I would like to explore and try my hand at solving in an open source commit as it would have made the architecture feel much more modular. Despite minor hickups, Axum proved to be a worthy and capable backend technology with a plethora of middleware available for serializing and deserializing data cleanly, creating a multithreaded database pool for efficient database querying across many devices, and creating a robust REST api. One final notable item that I had a great time using was ngrok server masking; I was able to utilize my own computer as a server for our mobile development by exposing the ports to a public url via ngrok. This was an exciting, fun, and fast way to develop that I intend to use in the future. All in all, I had a blast at RowdyHacks and greatly enjoyed developing in Rust!

Sam Cohen