ABS (Advanced Break-even System)

Inspiration

Many people are deep in debts and do not see anyway to get out of them. These could be student loans or mortgage payments or anything such. This app is designed to help them get out of it.

What it does

This app will give user a plan to get out of debt based on their income, expenses and total debts. This app will also include a chatbot which will answer any finance related questions the user may have.

How we built it

Frontend: React, TailwindCSS Backend: NodeJS Database: MongoDB, AWS CosmosDB Chatbot: Flowise architecture which includes Ollama Llama3 chat model, hugging face embeddings, Pinecone vectorDB. Finance Model: Fine-tuned Finance-specific LLM with data scraped from sources aligned with our problem statement. Friendly: Using this platform to deploy LLM and use the endpoint for fronted, making it simple to use the model with the web application.

Challenges we ran into

Structuring JSON data between front end to backend.
Hosting flowise to be able to hit the endpoint.

Accomplishments that we're proud of

Being able to build a fully functional web application.
Being able to leverage AI to be able to generate reports and also build a chatbot which answers user queries based on the domain.

What we learned

Using LLMs to work hand in hand with them to be able to work better, faster and more efficient.
Fine-tuning LLMs and SLMs specific to our use case.
Utilizing different hosting platforms to use and fine-tune models without too much hassle.