Inspiration

As rising seniors, we know we struggled a lot with learning how to succeed in high school from the start. Knowing that a lot of new highschoolers might also be in need of this sort of guidance, and that a well-trained and informed AI can provide real, data-driven insights, we decided to implement a RAG model to take user input and come up with custom feedback. The feedback helps in better managing time and also tracking activities, all while providing personalized feedback and counseling.

What it does

Users input their information in their profile tab after signing up/logging in, and then receive AI-powered feedback that suggests new extracurriculars, activities, competitions, and more that are tailored towards their experiences and interests. The suggestions update each time they update their profile, leading to a persistent, 24/7 available counselor.

How we built it

We started simple, developing a front-end webpage including just general resources that we had learned from our own experiences; for example, we added links and suggestions for different types of testing like APs, SAT, and ACT. We wanted to make the user experience more custom, so then we created a backend using Supabase and postgreSQL. There's a RAG system that takes the user's information as context to generate custom suggestions and insights, which are then relayed back to the user in their respective sections.

Challenges we ran into

We weren't too familiar with using PostgreSQL, but after doing some research, we decided it was the best option here. To make it work, we looked at a lot of documentation and online resources, meaning the code isn't perfect, only functional for now. One specific place where we struggled was in setting up login/logout, especially with email authentication, which we ended up having to remove.

Accomplishments that we're proud of

This was our first hackathon, so we're proud of managing to struggle through such a large project. We like the way the UI looks, since it's interactive and easy to use, and are proud of how the login/logout features are all functional, since this was initially a little tricky for us to set up with Supabase. Also, we're happy to have successfully adapted to using Render later since Vercel wasn't allowing us to use it, which we hadn't accounted for initially.

What we learned

We learned a little bit about PostgreSQL and working with Supabase. Managing sign up and sign in was a little difficult since we haven't done it before, but we enjoyed working through it and were happy when it worked. In addition, we worked on optimizing our RAG so that it's more efficient and uses less memory. We also learned an important lesson of starting small and only then expanding. When we initially tried to add in a lot of features at once, we struggled to stay organized and failed a lot, so we moved to just making small additions each time.

What's next for High School Compass

We still have a few small UI issues to fix on the front end. Beyond that, we'd like to find a way to make a smarter and even more intelligent RAG system that can give more unique and useful responses to the user. Another idea is to homogenize the way we categorize accomplishments and create a program to rank different awards/accomplishments systematically for all users. Google OAuth is another thing we'd like to add, since it's very common, and implementing email verification in login, which we had to disable due to rate limits. We'd also like to add in forgot password functionality, since that's lacking right now and we even forgot our own password once! A more general goal is to learn Postgre in the future and then polish our code here to ensure user safety and data security.

Share this project:

Updates

Submission history