Campus Budget Companion

Inspiration

Managing living expenses alongside tuition is one of the biggest challenges for college students. Without instant financial guidance, students frequently run into unexpected monthly deficits. We wanted to build a lightweight, accessible tool that bridges this gap by offering instant visual expense tracking paired with AI-driven, actionable budgeting advice tailored directly to student lifestyles.

What it does

Campus Budget Companion allows students to input their monthly income, rent/utilities, food, and entertainment expenses to immediately calculate their remaining monthly balance. With a single click, an integrated AI financial advisor analyzes their exact spending breakdown and generates personalized, real-time advice streamed token-by-token.

How we built it

Team: Mahogany Davis, Imri Carr, and Makyla Hudson
Frontend & Framework: Built in Python using Streamlit for an interactive, responsive web interface.
AI Engine & API: Integrated the Groq Cloud API utilizing high-speed open-source Large Language Models (openai/gpt-oss-120b).
Real-Time Streaming: Implemented Streamlit's st.write_stream generator pattern to stream AI responses live to the user interface.

Challenges we ran into

During development, we navigated and resolved several technical hurdles:

API Model Deprecations & 404 Errors: We initially hit groq.NotFoundError (Error Code 404: model_not_found) across deprecated model endpoints (such as llama-3.3-70b-versatile, llama3-8b-8192, and llama-3.1-8b-instant). We successfully resolved this by inspecting updated API models and migrating to openai/gpt-oss-120b.
String Formatting & Parameter Typos: Fixed runtime exceptions caused by minor string parameter typos (e.g., llama3-3-8b-8192).
Streamlit LaTeX Rendering Glitches: When currency values containing dollar signs (e.g., $500) were streamed, Streamlit interpreted them as LaTeX math formulas, stripping whitespace and rendering text in italics. We resolved this by escaping dollar signs (.replace("$", "\\$")) prior to output.

Accomplishments that we're proud of

Successfully connecting a fast open-source LLM through Groq API to a live Streamlit UI.
Implementing live response streaming so advice appears instantly as the AI generates it.
Building a complete, working financial tool tailored for college students from scratch during the hackathon.

What we learned

How to manage and debug third-party API deprecations and runtime errors under tight deadlines.
The nuances of Streamlit text parsing and string escaping for markdown and LaTeX formatting.
How to leverage open-source models for practical, real-world utility apps.