Adriana: The Fair Hire AI

• 

  Thumbnail

Inspiration

Imagine a world where every job applicant is given a fair and unbiased opportunity. In today's fast-paced and competitive hiring landscape, ensuring equity in interviews is more important than ever. Meet Adriana, the Fair Hire AI, an innovative solution designed to bring transparency and accountability into the hiring process.

What it does

Adriana analyzes interview recordings and generates a Fairness Score between 0 and 1 (0 very biased and unfair, 1 unbiased and fair), assessing the impartiality of the conversation. It goes further by using explainable AI to reveal how that score is derived after evaluating a recorded interview. But it doesn’t stop there. It breaks down the key elements that contributed to the score, offering a clear explanation that helps hiring teams understand where they may inadvertently introduce bias. Furthermore, it also supports continuous improvement in interviewing practices. With insights into fairness factors and specific feedback, hiring managers and recruiters can refine their approach, asking more equitable questions, focusing on candidate strengths, and conducting more inclusive interviews.

How we built it

Adriana's design is centered around providing a seamless user experience. The key design elements include:

• Natural Language Processing (NLP): It uses Google Gemini API to understand the context and intent behind interview transcripts, ensuring accurate and relevant response.
• Intuitive Interface: The app interface is designed to be user-friendly. Built using Flutterflow, its frontend is optimized for both iOS and Android platforms, ensuring a consistent experience across devices.
• Backend Integration: Powered by Google Vertex AI, the backend handles the docker image and deployment of the service.

Challenges we ran into

• Getting used to the Vertex AI environment. I had many issues with building the docker image and deploying service. There aren’t mainly references out there.
• Speech redering in different browsers. Some browsers use a male voice instead of female, especially in desktop browsers.

Accomplishments that we're proud of

• Getting conversant with the Gemini API and Vertex AI environments.
• AI for good: Contributing to transparency and accountability into the hiring process.

What we learned

• A lot about Gemini API and Vertex AI as well as other options for GenAI and MLOps.

What's next for Adriana: The Fair Hire AI

Over time, Adriana becomes an essential partner in the journey toward fairer and more effective hiring.

Built With

• docker
• fastapi
• gcp
• gemini
• javascript
• poetry
• python