Resumescore.AI

Inspiration

Being a mix of computer science and data science students we know the pain point of applying for internships. One of us recently got an interview after applying to over 200 positions and applications. He expressed how relieved was to get the first interview that truly made him happy. I kept mentioning if he could have found this application earlier maybe he would have saved time and stress. That's when the idea struck. What if we can truly save that time and stress. Then Resumescore.AI was born.

What it does

It takes in your resume and using AI we match it with job postings from websites like indeed and Linkedin. Taking all this in as parameters the AI model gives out a match score. This score indicates the likelihood of you getting an interview. For now it gives you the top 3 choices but we want to be able to give more in later versions.

How we built it

We used a variety of technologies, primarily focusing on Python’s datascraping and data analysis libraries, mongoDB to create our database of job descriptions, and then React JS to build or frontend. The key way we created our similarity scoring system was by vectorizing the pdf the user would send and then vectorizing each individual job description and then using cosine similarity to find which jobs description best match with the user’s resume. From there we pulled our job datasets using webscarping (selinium library in Python). After inputting thisw database into mongoDB, we then built up our React app to create an engaging and useful user experience.

Challenges we ran into

Throughout our development process we many times got stuck on how best to make our program run efficiently. This involved figuring out how we could best build up our database for job descriptions by using APIs vs building a web-scraping tool. Another point we got stuck with was how we could build our own API so that the other developers of the team could easily test out our application. This involved doing heavy research on best technologies to use (e.g. Flask). Finally, as we were building up our application we tried to put a heavy emphasis on user experience which led us towards problems relating to website functionality.

Accomplishments that we're proud of

figuring out how to working with mongDB and building a solution that could have some real impact.

What we learned

We learned that making an AI software isn't that hard at all.

What's next for Resumescore.AI

We are looking at expanding the platform in such a way that we apply for these positions for you. So your even more hands off with internship hunting. Saving your time and using it for honing in your skills.