Inspiration
All over the United States, high schoolers face the dawning challenge of getting into prestigious colleges/universities helping decide how they advance in the workforce and life in general. But they can't reach the next step in their life without advice regarding how to reach it. Currently, college counselors monitor a students progress throughout high school, specifically how they advance in courses and their overall performance in school, making a fairly accurate guess on which school the student should apply to. The only problem is that a lot of the time, their advice is not the most accurate. This is why we created CounselorGPT.
What it does
CounselorGPT measures academic courses and GPA to determine the best college for high schoolers over the four high school years. The GPT processes the data as the high schooler select their courses in the dropbox of the model. It uses OpenAI API to calculate a dataset of different colleges' GPA and number of AP courses prerequisites to determine the best college for the high schooler.
How we built it
The GUI of the project was made. Then functionality was added to each part of the GUI such as inputting course loads and GPA as well as saving this information. OpenAI was introduced and the previous information was fed to it.
Challenges we ran into
We encountered many bugs and incompatibilities, especially concerning the API keys, and implementing GPT-40. Many of the APIs aren't compatible with our model, and we continuously encountered several errors. As well as this, it was an extremely difficult endeavor to procure datasets for this project. This was because many public datasets are limited in their knowledge of college admissions, especially high school courses, and course pathways. So, we manually web scraped to procure datasets that would fit our style and product, and our unwillingness to compromise for less, ultimately led to our success.
Accomplishments that we're proud of
We are extremely proud of our UI and website such as the successful initiation of the code and the website design. The OpenAI API also worked perfectly and there were nearly no bugs, which we created a working and consistent model. Furthermore, the website is also working for its publicity and visibility.
What we learned
We learned that some APIs and datasets are compatible, but some aren't compatible with our model. We also learned about making a working website for our AI model.
What's next for CounselorGPT
We intend to expand the boundaries for our AI model by including SAT scores and extracurriculars from high schoolers. This helps the accuracy of our CounselorGPT.
Built With
- cohere
- css
- html
- javascript
- openai
- python


Log in or sign up for Devpost to join the conversation.