inteliBudget

View of the API Keys in the Function App Resource

Inspiration

When brainstorming ideas for this project, we thought that building a tool to help users gauge their budget with the help of AI was a pretty nice and practical concept. It would give another voice to individuals who may not have another avenue to receive advice besides the bank.

What it does

It generates a budget for the user based on the transactions and savings in their bank account.

How we built it

This application was intended to be built using Vue.js for frontend elements, Azure OpenAI o-3 mini (Global Standard) for orchestration, Azure Functions to host our application, Azure Application Insights to monitor for issues and errors, Azure Storage Accounts to store user data, Azure Key Vault to securely store API keys, Nvidia Nemotron to generate the blurb describing the users financial situation, CapitalOne's Nessie API to pull all their financial information (such as transactions and savings). In the final version we could finish the backend implementation but we our front end gives a picture of how the app should function.

Challenges we ran into

The orchestration of the various AI agents proved particularly difficult to put together. Additionally, as this was our first time using Nemotron, we were unfamiliar with its deployment and functions, causing us to encounter a multitude of bugs. Additionally, as we were not very familiar with database integration, we encountered a lot of implementation bugs and errors, taking up most of the hackathon period.

Accomplishments that we're proud of

As a team, we developed a well-thought-out plan of action to the hackathon with ideas on the implementation, orchestration, and database integration. Given that this was the first time some of us even heard of these terms, it was nice to be able to learn new things as a team in such a short window.

What we learned

We learned Azure basics, such as how Azure works behind the scenes, creating resources, allocating space, linking different resources, adding components to each resources, and trying to deploy the app function. Additionally, we learned about Nemotron and what is does and how to implement it as well as how Nessie works.

What's next for inteliBudget

As one of the team members has an upcoming interview on cloud engineering, this project will continue to be worked on - ideally in the cloud deployment stage, so to give the team member practice on deployment and to finally get our project running on the cloud.