Inspiration
We were inspired from the themes provided from the hackathon. We decided to focus more on student problems and solutions to help peers around us. We looked at frequent problems that we personally encountered as students and tried to resolve that problem.
What it does
We decided to focus on breaking down syllabuses provided by courses. Our web application helped breakdown the syllabus to ease referencing important dates. It down to key things like professor name, key dates, TA hours and more.
How we built it
We split up different tasks into research, UI, and coding the application.
Challenges we ran into
A challenge we ran into was an issue with our hardware. Our original plan was to have our LLM run locally with a small model called Falcon 7B. However, our hardware could not accommodate, our laptops were short about 2 GB of RAM which caused a drastic decrease in performance. Our alternative was to pivot into using OpenAI API which offered their own hosted model, meaning we did not have to worry about our hardware.
Accomplishments that we're proud of
We are proud to have officially created our first AI project across our whole team. This project consists of a new tech stack that none of our team has worked with before. We feel very accomplished having finished this project.
What we learned
Prior to this hackathon, we did not have AI experience. But in order to create our product we decided to learn about LLMs. Our idea was very solid, but were unsure how to execute it by doing further research we decided that using OpenAI's GPT3.5 turbo model would be the best solution for us.
What's next for Syllab-US
In the future, we hope develop more functionalities specifically things like project prompting. The AI would be looking through any project descriptions and figure out what tasks we should start off with or even help us through the ideation phase of projects. We also saw this problem occur when we were developing the idea for our application. It was difficult to find a place to start, but once we got a general idea of how we wanted to start everything went much smoother. Some other things we hope to add in the future are things like importing course dates into personal calendars and finding information on professors (course materials and projects prior to enrollment in the course).
Built With
- adobe-illustrator
- figma
- github
- gpt3.5
- mysql
- next-auth
- nextjs
- openai
- prisma
- python
- typescript
- vscode
Log in or sign up for Devpost to join the conversation.