Our Inspiration We noticed that college students, both graduate level and undergraduate, can struggle to secure an internship or a job. We looked into why and we came to the conclusion that students are struggling with their interview process. We decided to build PrepSTAR which seeks to help students with the interview process and utilizes the STAR method. Using the STAR method, PrepSTAR gives feedback based on the students response allowing for students to improve on their weaknesses.
Our Goal* PrepSTAR helps students practice interviewing before they actually go to interview for their dream job or internship.
Our Resources We built PrepSTAR utilizing Vapi API, OpenAI GPT4.0 Mini, React, Jsx, CSS, Javascript, and Python frameworks.
Challenges We ran into some challenges, such as when we where prompting the AI to give better responses as well as more detailed feedback. The AI had issues with interrupting the user who was speaking to it so we had to adjust the AI model and fine tune it to help reduce the frequency of this issue. Additionally we ran into other issues such as the AI not knowing what to look for in the conversation, resulting in us using prompt engineering in order for the AI to parse through the conversation better. After the prompt engineering the AI delivered better responses since it better analyzed the exchange.
Results We're proud that the website and AI works as it is intended and delivered results we feel can help students be better prepared for interviews. Most importantly, we are quite satisfied with the category breakdown, where the AI coach gives the user a numerical score for each category.
What we learned We learned how to integrate the Vapi API with the OpenAI LLM model GPT4.0 Mini and we also learned how to integrate React with Javascript and other frameworks in order to deliver PrepSTAR.
The Future What's next for PrepSTAR is to further the prompt engineering of the AI so that it can deliver even better results and eventually take PrepSTAR to the next level by delivering more features and feedback for the user.
Built With
- css
- gpt4.0
- javascript
- jsx
- python
- react
- vapi
Log in or sign up for Devpost to join the conversation.