FairInterview | Devpost

PITCH PRESENTATION LINK: https://youtu.be/fiyuF7yo748 PRODUCT DEMO LINK: https://www.youtube.com/watch?v=iiIz53k6ngA

Inspiration

As female visible minorities who have worked in large corporations, recruitment diversity and inclusion have always affected us. Despite many companies saying they take it seriously, countless studies and research have shown diversity recruitment has remained a challenge. Since COVID-19 lockdowns have shown us that video call interviews are essential and the trend is likely to proceed. We think a blind hiring interview tool will be in high demand.

What it does

FairInterview is a comprehensive blind hiring tool that removes bias factors during live video interviews. It preserves candidates’ personality by keeping real-time facial expressions and voice tunes. The main features include a face swap, voice changer, and auto-hiding name.

How we built it

We started by talking to friends who are doing recruitment around us, and figured out this might be a general pain point. Next, we did thorough market research about the unmet need, pain point, opportunity and potential competitors. Then, our team, with backgrounds in AI & Machine Learning, Design and Marketing, began to work on the product concept and kept receiving feedback from the user test. Finally, we finalized our MVP product with positive feedbacks from the potential users.

Challenges we ran into

The time limitation is definitely the biggest challenge. We are working on UK/Canada time zones, which makes it more challenging. The second one is asking for quality feedback. Because we are targeting hiring managers that can make purchase decisions. It requires a very selective sample. Finally, it’s the technical side. Obviously it’s impossible to build a GAN model in two days. Therefore, we found a work-around to utilize existing GAN technologies first from apps like Avatarify.

Accomplishments that we're proud of

We did it! And happily.

What we learned

We learned a lot about the importance of talking to users and hearing the different perspectives; not everyone was an instant fan of the product’s features and we learned how to take constructive feedback but also know when a test user wasn’t necessarily a target user.