Prepr

A novel approach to preparing for your future interviews.

Comment 1

Inspiration

As 3 computer science students, we understand the struggles of the behavioral interview for students and how it can be a large hurdle when being introduced to the work force. We developed this tool as a way for students to have infinite unique practice with an artificial interviewer so that

What it does

Prepr is a novel approach to interview preparation. Using prepr, users can have a mock behavioral interview in real time with an artificial interviewer. The interviewer will ask general behavioural questions that build upon the answers of the users. After the interview, the user is then presented with a set of analytics that show how well they performed in the interview. The analytics presented are percentage of time the user has kept eye contact with the interviewer, the amount of general key words that they used in their answer, and the number of filler words that they used such as "um" or "uh".

How we built it

The frontend was build using react native. OpenCV and mediapipe were utilized for the eye tracking and Flask was used to write our RESTful API. The conversation generation was done using the Kaldi and Vosk speech AI models as well as openAI's API.

Challenges we ran into

The first main challenge we faced was being able to run the eye tracking model at the same time as the speech recognition and feedback model because they both require infinite runtime. This problem was quickly solved using multithreading.

Accomplishments that we're proud of

We built quite an impressive application in the span of only 36 hours that has real potential to become a large compant. This application used 5 largely different AI models, and uses them to create a user experience that is unique 100% of the time and solves an extremely imminent problem.

What's next for Prepr

We believe that Prepr is the future for interview preparation and we are prepared to take it all the way. After this hackathon we plan on continuing development and eventually applying to Y Combinator in hopes of getting funding for our startup.

Built With

Share this project:

Updates