The members of our team are quickly approaching the end of their degrees, and will soon be sent off into the professional job market. Taking into consideration current affairs such as the poor state of the economy, and an astronomical amount of layoffs in the tech industry, many soon-to-be graduates (including ourselves) are scrambling in hopes to find a secure job.
This is the main reason we decided to develop Jane: to help these students hone their interviewing abilities, master the art of rhetoric, and annihilate any coding challenge that steps in their path to success.
What it does
This web app uses a tailored Artificial Intelligence model to create natural-ish flowing mock interviews. It provides sample behavioral and technical questions, gives advice and constructive criticism, and highlights the user's strengths and weaknesses when it comes to the interviewing process in order to help them improve their likelihood of securing a good job.
How we built it
The core foundation of our web application is built upon Next.js, a react meta-framework. We are using the Web Audio API to record and encode audio information and stream it to our back-end using WebSockets. This data is then parsed by Google's Speech-to-Text API and analyzed by a version of GPT-3 tailored toward simulating mock interviews. The data that is returned is then converted to binary audio data to be read out in a Text-to-Speech voice thanks to Google's Text-to-Speech API.
Challenges we ran into
- The challenge that impacted us the most was the initial brainstorming phase. It was very difficult to come to a common consensus on our project. We came up with many ideas, most of which we nit-picked to the grave.
- Another important challenge was understanding the various methods of recording, encoding, streaming, and analyzing speech data. This was definitely the most nerve-wracking part of the project.
Accomplishments that we're proud of
We are very proud of the user-centered design and the overall interface and interactions within the platform. We are also proud of the amount of effort we put into the speech-to-text and text-to-speech aspects of the application. It was a daunting task, and none of us had any prior experience, but we managed to overcome these hurdles.
What we learned
We learned the extent to which technology has advanced in such a short amount of time. The feat that we completed would have not been feasible to do in a week maybe 5 years ago, let alone 24 hours. We were able to explore the vast quantities of interesting APIs and services offered. Finally, we learned that by working as a team, dividing and conquering our work, and properly planning and ideating, we can succeed in completing challenges we may not have considered possible.
What's next for StackingTables
Our goal is to hopefully pursue this idea in the future (hopefully with a more stable foundation), but for now, our main objective is to complete our degrees :)