InterviewBot

Logo
Interview Flow
Landing page
sample question set
Domain name

Inspiration

Behavioral Interviews are a very crucial part of the interviewing process. They help the company judge whether the candidate is motivated, is able to learn new technologies and believes in the company's values. However to prepare for behavioral interviews, candidates have to practice with their peers or practice in front of a mirror. Practicing with peers is helpful but your peers might not be qualified enough to provide quality feedback after the interview. *ChatGPT * has the ability to generate a high quality feedback when prompted correctly so our team decided of leveraging that and creating a chatbot around behavioral interviews.

What it does

This is a bot which conducts a behavioral interview and provides feedback to the user based on how optimistic and motivated the user is, and how eager he is to learn new technologies and skills. The user is given a question, his response is recorded from his mic and sent to GPT- api for critical feedback.

How we built it

We used Reactjs and TailwindCSS to create a frontend for the chat and the report page. We implemented speech to text functionality using react libraries. The questions and answers are sent to gpt-api to be evaluated . We used the questions, responses and the feedback to generate a report after the interview is over.

Challenges we ran into

Determining which gpt model to use for our use case (we decided to go with davinci-03).
Prompting gpt api correctly to get the desired response.
Handling a lot of asynchronous calls.

Accomplishments that we're proud of

We were able to successfully prompt, get and display feedback from the GPT API.
We were able to implement speech to text functionality.
We were to mimic a complete interview flow.
Created a good MVP in 36 hours.

What we learned

We learnt how to prompt chatgpt api for our specific use case and also learnt how to implement various libraries in React JS. Also learnt how to implement different react hooks. We also learnt how to use face api to implement emotion detection for the future.

What's next for InterviewBot

We want to integrate the video view with an emotion detector and also reaction detection so that if a person is getting diverted he will be notified in realtime. This will enable a better feedback since body posture and facial expressions would also be included in the feedback.
GPT can also be used to ask good follow up questions, this is a feature we will work on in the future.