CareerConnect

Inspiration

We were inspired to build this app due to similar experiences that we have had in our current (and ongoing) search for internships. Internship hunt is the single-handedly one of the biggest struggles that every university student can relate to. And the larger your program, the larger the competition. Unfortunately for us, not only are we competing with hundreds of other students against each posting but we also need to ‘compete’ with the sudden downturn in the economy that has restricted even larger companies from hiring their usual number of interns, with most companies going on a 'hiring-freeze'. We wanted to use the wonderful tool of artificial intelligence, to make the process of going from resume to interview, a little bit smoother.

What it does

CareerConnect is a web app built using AI generation to help your internship hunt, by improving the most important thing an internship application has; your resume. We've all heard of the (very) famous saying 'Recruiters only look at your resume for 6 seconds!' , but we could agree that in today's time, mostly your resume isnt even looked at by a recruiter before you reach a second or third step in to the application process. The app will take your resume and the job description that you are interested in applying for and optimize your resume so that your matching experiences can also be written (and read!) as equivalent experiences. This allows you to stand out amongst others as our tool tries to best match your experiences to the provided job description, to let the recruiter know that you are a perfect fit for the role.

How we built it

We used the cohere generate model and openAI text generation model, and trained it using some publicly available datasets to be able to achieve more accurate results. The front end is powered by Next js, with Node.js and Express for our backend. The resume is initially parsed into a json object and then passed onto the fine-tuned model which generates a result.

Challenges we ran into

This was the first time we worked directly with AI, and ran into some issues with the fine-tuning and training process for the models. However the online documentation for the AI models we used was in quite some detail and although it was harder to understand, it took us some trial and error but we managed to work our way through it.

Resume parsing was also an issue, which we were able to complete by adding some minor changes to an open-source resume-parser library, available in javascript.

Accomplishments that we're proud of

We are happy that whilst we walked into this project with little knowledge about AI models and natural language processing, we not only learned something new that will benefit us in our future projects but that we were able to complete most of what we envisioned and achieve the main functionality for our application.

What we learned

We learned quite a lot about how to use artificial intelligence models and how we can integrate them in a full stack web app. We were always of the impression that training and fine-tuning AI models would be an extremely difficult task, but are happy that we were able to study about it and implement it in our project. As the world advances into implementing AI technology everyday, we are happy that we were able to take our first steps into building something with artificial intelligence at SheHacks!

What's next for CareerConnect

The vision for CareerConnect is such that we aim to expand it so that users are able to sign up for accounts, and can add some personal information about them. This data can be used to search for internships, specific to their sector and interests whilst generating a resume and/or cover letter for the applications that they choose to apply for. As students, we understand the struggles that come with finding an internship, and we want to be able to create a tool that we and others like us can benefit from and make our college years, a little easier :)