Unemployment No Longer

Been looking for a job to no avail? Hoping to have more practice before any interview? Here's UNL! Using UNL you can have the instructiveness of a coach without the inconvenience of leaving your home

Comment

Inspiration:

Unemployed No Longer (UNL) began as a discussion on the terrible job market and how companies lack the capability to get back to students and provide feedback on interviews. UNL aims to give students this feedback, allowing them to ace the next interview.

What it does:

UNL gives you an interview prompt based on inputs; taking in an existing job ad you are trying to apply for (if you have one) as well as prompt difficulty and category. The software then tracks your posture and eyes during the interview ensuring you are maintaining correct posture and eye contact, and records the interview audio. The audio is then processed and your response to the prompt is evaluated, informing you what you did well and how you can improve.

Software Used:

In the development of UNL, we utilized the following software: Frameworks: Vite + Reactjs (Frontend) and FastAPI (Backend). Libraries: Media Pipe, OpenAI, Librosa, Numpy, Groq, Whisper, Recharts, MatPlotLib Programming Languages: Python, JavaScript Web development: HTML, CSS. Distributed Version Control System: Git.

How it works:

  • The front end takes in the user's job ad, category, and desired prompt difficulty. If no job ad is given, it will generate a prompt from the backend database containing 250 pre-loaded prompts.
  • The software will then move to a countdown, giving the user a 3 second countdown before the interview begins.
  • A 30 second think phase will initiate, giving the user 30 seconds to prepare a response to the prompt
  • The 90 second response phase follows, recording the interviewee's audio, posture, and eye data.
  • The data is then uploaded to the backend using REST API calls where it analyzes tone, pitch, speaking rate, and overall response using GroqAI
  • The response is then uploaded to a JSON file, where it is read from the frontend and displayed on a Results Page
  • The user can then start a new interview and download results from this page.

Challenges we ran into:

  • One main challenge was implementing the review section onto the results page and ensuring smooth and constant communication between the frontend and backend
  • Another challenge was uploading the audio to the backend for further processing
  • We also faced challenges with prompt formatting for groq to give us a desired output
  • We also faced challenges with the CSS, trying to make it as user friendly as possible
  • Lastly, we faced a challenge when formatting the JSON file for further reading

Accomplishments that we're proud of:

We’re proud that we were able to combine our individual strengths to build a fully functional prototype of UNL without any major glitches or bugs. Within a limited timeframe, we successfully integrated real-time video analysis, audio transcription, and AI-generated feedback into one system. Most importantly, we created a product that isn’t just a technical demo but a tool that we would personally use and that anyone preparing for a job interview online could benefit from.

What we learned:

  • We learned how to collaborate in a dev environment using github
  • We learned how to process audio and analyze it
  • We learned how to implement AI into a project

What's next for Unemployment No Longer

We hope to create a more polished, professional version that is Git-Hub ready and has more accessibility features like converting text to speech for individuals unable to read and track past scores to show overtime growth and improvement for users.

Built With

Share this project:

Updates