Inspiration
As students currently in recruiting season, we know the stresses associated with interview prep. Inspired by these struggles and Year of AI, we decided to create a tool that simultaneously benefits helps other students in tech just like us while harnessing the unique and powerful capabilities of existing LLMs and computer vision.
What it does
InterviewMania conducts a brief mock behavioral interview, using computer vision to track eye contact while using LLMs to both create questions and assess responses. At the end, a user can choose to save a transcript of their mock interview.
How we built it
After deciding on our idea and what exact features/look we wanted to implement, we divided work between backend and frontend development, with one member helping out on both sides. After getting basic features working on both ends, we combined our work into one product using flask.
Challenges we ran into
For frontend, we found it quite difficult to implement 3D space into our work. We ultimately decided to use a solution in between our 3D goal and the convention of typical websites where the actual function of our product was based in conventional 2D. On the backend side, we ran into hiccups with using free APIs, as well as challenges with developing our algorithms. Overall, we also encountered issues bringing together different ends of our development with flask. Because of our approach to computer vision and the look of our site, it was quite difficult to format a cohesive body.
Accomplishments that we're proud of
We're proud of the final functionality of InterviewMania as well as the cool look of our user interface.
What we learned
There are a lot of intricacies involved with combining different languages and sources, but the Internet has tons of resources that will help you learn.
What's next for InterviewMania
In the future, we hope to add an advanced body language analysis (also using computer vision), further develop the feedback of our interview, and train the existing LLMs we used on more specific data to provide better questions and response analysis.
Log in or sign up for Devpost to join the conversation.