CCCompass

CCCompass Logo

Inspiration

While top-tier universities enjoy smooth, integrated academic platforms, CC students are often left navigating outdated catalogs and scattered information sources. Born from the frustration of outdated catalogs and fragmented info, we built CCCompass to give Community College students the elite academic tools they deserve.

What it does

CCCompass is a website designed to cover all community college students in the United States(Demo on SBCC), providing services such as Catalog, Smart scheduler, AI course recommendations, classmate group finder, alumni resource links, and graduation progress track.

How we built it

Use MongoDB as a non-sql data base, we imported the current and past - future 5 years data from SBCC official website, we built a framework that implements functions through Rapid Fire api to get an on-time AI response. Also, from the past graduation results, we trained our own Random Forest regression model, which has a 0.9 R-squared score to suggest future coursework. We use Live Data Technologies to select Alumni data for our Alumni Networking part.

Challenges we ran into

Integrating high-density data streams from smart schedulers to alumni hubs into one dashboard caused massive layout shifts. We also faced significant friction connecting our initial Python-based API logic to the live framework. This forced us to reimplement several core functions from scratch to ensure system compatibility. Due to the lack of direct access to official university databases, we trained our model on a synthetic dataset. This led to overfitting. We recognize this as a critical area for improvement as we seek more diverse data sources.

Accomplishments that we're proud of

Our proudest achievement is the creation of an all-in-one data ecosystem that bridges the information gap for CC students. We engineered a centralized platform that integrates real-time catalogs, professor ratings, and alumni networks into a single, intuitive interface. By combining this data engine with our predictive model, we’ve replaced outdated pages with a high-tech academic GPS, ensuring every student has equal access to elite-level planning resources.

What we learned

One of our most significant breakthroughs during the development process was discovering that AI-assisted coding is far more efficient when driven by precise JSON schemas rather than abstract natural language descriptions. We found that by strictly defining the data structure first, the AI could generate much more consistent and reliable code, significantly reducing the hallucination and logic errors often found in complex framework implementations.

What's next for CCCompass

Moving forward, we plan to refine our predictive model using Regularization and Cross-validation to eliminate overfitting and ensure robust performance across diverse academic backgrounds. By leveraging our structured MongoDB database, we aim to transition into a RAG-powered AI advisor that provides every student with a personalized, data-backed, and 24/7 accessible roadmap to graduation.