Inspiration
Beginning university is a difficult change for anyone, and many new students decide to enter as Undecided Majors. The purpose of our website is to recommend majors and student organizations for Panthers if they are in a position where they need guidance.
What it does
This website operates on a Q&A premise. Users input data and answer a few questions, then click to receive either career or networking recommendations. We also began implementation on giving all recommended courses of a major based on the FIU catalog.
How we built it
We employed Google's generative AI model Gemini (used through API calls) to generate recommended responses focusing on FIU majors and official student organizations. Gemini is called in two prompts, one for majors and the other for student orgs. The prompts include lengthy information on all majors and student bodies at FIU.
We utilized JavaScript and Node.js to employ dynamic functionalities to our website, created via HTML and CSS documents. We also have a login system that, in the backend, saves user information in a database, which was created using Python.
Moreover, we used JavaScript to parse the PDF of all FIU courses. Gemini parses through this document to find classes related to the major the user prefers, however this feature is still in development.
Challenges we ran into
Our idea was the first challenge, it took us some time to hone in on the nature of our project. We also had to drop numerous features we had in mind at first. Initally, we wanted to web-scrape RateMyProfessor to recommend the high-rated professors to students, but their TOS does not allow this, so we had to move on without this functionality.
We ran into numerous challenges while coding. Initially, we tried two other AI models (ChatGPT 4 and Ollama) before we found Gemini, which worked excellently for our project. In general, there were many issues with error handling and debugging. We also needed to figure out how to merge different tech stacks like static HTML files, node.js, and Python FLASK (for the DB).
Scraping the courses from the FIU PDF was difficult to accomplish. Gathering the data for organizations and majors at FIU was also a tall task. We also needed to tailor the Gemini prompt in a specific format after numerous tries to make the parsing more consistent.
Accomplishments that we're proud of
We're proud of the look of the website, its sleek design, and attractive UI. It's intuitive to navigate through and our career and networking recommendation features are quite accurate.
What we learned
Throughout this project, we gained experience in working together as a team of 4 in our very first hackathon. Moreover, we learned many new technical skills, such as employing an AI model to generate recommendations based on user input through API calls, database management, web-scraping, successfully connecting the front-end with the back-end, and the importance of communicating with each other to solve issues.
What's next for Panther Vision
Ideally, we would love to include professor recommendations for all students, however, this feature is indefinitely on hold. Realistically, we would get the course-recommendation functionality to fully operate and we would love to include a generated flow-chat for the major, offering future career goals alongside networking advice. Our site would be a one-stop-shop for Panthers looking for success.
Log in or sign up for Devpost to join the conversation.