Inspiration
Going over the sponsor challenges, what inspired us was the real-world problem OneEthos's challenge was trying to address and the proposed projects that would aim to solve them. It was vague enough to start a conversation but focused enough to realize the real-world impact of such a project. It got us talking about our life experiences and technology that we would want to solve those problems.
What it does
Our web application uses Plaid's API to collect transaction records to establish inflow and outflow for classification and analysis. Using that data, our web app labels data into various financial categories to create a clear and simple picture of one's financial situation. Labelling the data additionally aids our chosen LLM, Gemini 2.5 Flash, to target areas of spending with targeted advice to create a plan that could actually work in practice.
How we built it
First, we discussed and planned out the input and output and what kind of software components we would need to achieve those goals. We then picked out our frameworks based on familiarity and our goals and set up our environment for easy collaboration. The first day and morning of the next was used to build out the skeleton of our project, refining our environment, introducing new libraries, and engaging in better GitHub practices. Once development on our features began, we tried to work in parallel, intersecting on each other's tasks when relevant until features reached a satisfactory point for our timeline.
Challenges we ran into
Some of the challenges we ran into were developing a rhythm in feature development across frontend and backend, keeping pace with our goals, cutting down on the number of planned features to keep on track for at least a minimum viable product, and learning the frameworks we had chosen and the best ways to use them.
Accomplishments that we're proud of
Getting a minimum viable product ready for submission, our robust project planning and roadmap, and the fact that everyone contributed substantially at every stage of the project. Learning new skillsets and ways to apply AI and LLMs.
What we learned
We learned how to interact with APIs, what to look for and expect. Best practices for GitHub with APIs and whilst collaborating on the same blocks of code. We learned new frameworks and databases, learning their strengths and weaknesses firsthand.
What's next for Budget Ally
We have a decent list of features that we discussed and cut from our plans, even ones that we originally planned on implementing, and the project still has plenty of potential acting as a springboard to learning more about APIs, LLMs, and databases. Budget Ally will likely continue to be built upon in the name of learning.
Log in or sign up for Devpost to join the conversation.