Inspiration

We took our inspiration from Duolingo, which is able to make something as hard as language learning super compact and even fun at times. Since everyone in the American public school system will have to take the SAT at some point, we wanted to make it easier for people who wanted to prep for the SAT as well.

What it does

Our web app generates a much smaller version of the SAT using official college board SAT questions in order to compact SAT prep to 10-30 minutes a day depending on your preferences. We also query data from openai's chatGPT3.5 turbo(LLM) and plan to use user validation to make sure these questions are valid.

How we built it

We first scraped 2000 questions from the official college board practice SAT website. We also queried 100 reading and writing questions as well as 100 math questions from openAI's chatgpt3.5 turbo(LLM). We then added all of our questions into the MongoDB Atlas database where we used our web app built on React and Node.js to draw questions from. We kept good version control with good use of github. We also made sure to use Auth0 to facilitate secure user signups.

Challenges we ran into

One of the most annoying challenge I(Jason) ran into was that devpost did not save the project story, so I am rewriting it. Another challenge was that we data we scraped from the College Board's Official website was in a very messy html format. It was also really hard to get the front end connected with the database and the backend as well because we had most of the frontend finished before we even started on the backend. Another thing we ran into was that 2 of our teammates spent 2+ hours trying to find a free LLM API to call but we did not find anything good.

Accomplishments that we're proud of

Some accomplishments we are proud of that we were able to get a MVP in the time frame we had. We honestly did not think that we would be able to connect everything, even though we did scrap a few good ideas just to get it working I am very proud of what my team and I accomplished this Hackathon.

What we learned

While building the project we learned a lot about using LLMs since we did a lot of research on them while trying to find a free version. We also learned a lot more about authentication with Auth0 as most of us only had experience working with oauth. We also learned how to use MogoDB as we only really had experience using prosgresSQL and mySQL databases.

What's next for Banana Central

Some things that we wanted to implement before the deadline but did not get the chance to was to have a report feature for questions that would enable us to validate AI generated questions which would allow our app to scale as we would have unlimited questions. Another thing we wanted to add was a friends list as well as a profile page that would tell you your streak of how many days in a row you did a lesson on. Some more t

Built With

Share this project:

Updates