# Inspiration
I am a highly competitive individual, and one of my biggest passions is continually seeking ways to challenge and improve myself. However, finding appropriate platforms for this pursuit is often a broken, frustrating, and complicated process.

This personal frustration led directly to the idea for this project. After wasting significant time and missing a key deadline, I recognized a common pain point faced by students nationwide. CompComp AI was created to fill that gap—giving students everywhere a seamless way to bypass disorganized data, avoid missed deadlines, and jump straight into action.

# What it does
CompComp AI is a software platform that uses a matching algorithm to align a student’s detailed profile with current national academic competitions. Users provide defined demographics (age, grade, location), set their participation preferences (online or in-person), and enter their areas of interest. When users submit the form, the platform dynamically generates a grid of competition entries pulled in real time from a structured backend database. Each card shows the competition's name, format, location, and submission deadlines. A sidebar consistently summarizes the user's selections and profile.

# How we built it
We built the application using a frontend stack of vanilla HTML5 and CSS3, leveraging grid and flexbox for a responsive dark-themed interface, and asynchronous JavaScript (ES6) for real-time interactivity. For the backend, we implemented a serverless PostgreSQL database hosted on Supabase to store and retrieve competition data through RESTful API calls. We used Cursor, an AI-assisted development environment, to accelerate coding logic, manage UI form state, and handle data synchronization between the user-facing client and the backend database.

# Challenges we ran into
Before this hackathon, I had not integrated AI tools in software projects, and I faced difficulties with the AI’s handling of competition search logic. Initial implementations provided inconsistent filtering and output when processing form input, posing a significant technical learning challenge for AI-based query workflows.

When users entered form data, variations in input phrasing or notation disrupted the script’s matching process, occasionally producing null or irrelevant results from the Supabase-backed data. To address this, we re-engineered the evaluation logic by switching from fixed-text comparisons to a more flexible, case-insensitive string-matching algorithm, improving reliability and ensuring that students received relevant results despite minor input differences.

# Accomplishments that we're proud of
Beyond the project's technical accomplishments, I am particularly proud of how much I learned about AI and its future potential. Pushing myself to implement live backend query workflows—and realizing how AI is shaping the future of software engineering—marked a significant personal breakthrough. Turning a real challenge into a functional prototype that addresses students' needs is something I am deeply proud of.

# What we learned
This project taught me about both the strengths and risks of AI. I saw firsthand the immense productivity gains and rapid prototyping AI offers, but also the matching inaccuracies and the need for precise, human-written guardrails.

# What's next for CompComp AI
Next, we plan to improve the AI, deepen our understanding of it, and apply new knowledge to CompComp AI. We aim to upgrade from basic keyword filtering to semantic AI matching by integrating a serverless Large Language Model API. This will let the platform understand a student's natural-language biography and pair them with competitions, even if exact keywords differ. We also plan to expand the use of AI and enhance the user experience for students.