Talentide

Inspiration

We've been to a lot of job fair ourselves, so we know firsthand how busy it is. There's only time to talk to a few employers, so it's important to have productive conversations. We wanted to have a site to make it easier for students and employers to get the most out of their job fairs.

What it does

Our site makes use of that time by ranking the top employers that you match with and their open positions. It also shows you how many other students matched with them, so you can gauge how busy it will be and build a schedule. It's useful for employers, too; they can see how many students are matching with their company and what they're matching on.

How we built it

Our front-end is VueJS with HTML/CSS, and our back-end is python. We parsed the pdf resumes and job descriptions with the pdfplumber and nltk python packages. We first tried to use HDP (hierarchical Dirichlet process), but we decided TF-IDF would be a better way to go. We used TF-IDF over a KNN model to find the best matches between resumes and job descriptions. We focused on hard skills, extracting the keywords from the resumes and job descriptions and weighting them based on how many times they show up, and using machine learning models to fit them together. Finally, we used Cloudflare to join the back- and front-end together.

Challenges we ran into

Cloudflare was hard to work with, particularly when moving PDFs between front- and back-end. We weren't able to get the upload page completely working, so the two ends are not completely connected.

Accomplishments that we're proud of

We were able to clean the data and develop the model fairly well, and we learned how to use OAuth in our sites! We also learned a lot about full-stack and Cloudflare, despite not being able to get it to work in the end. We're also proud of the idea and hope to work on it later.

What we learned

We learned that joining together back- and front-end is a much bigger challenge than we anticipated.. and that we should have tried out different frameworks.

What's next for Talentide

We hope to connect our site to our back-end and get it fully functional so some version of it could maybe be used by the UMD career center! We know many students and recruiters would really appreciate it.

Built With

• ai
• backend
• cloudflare
• css
• css3
• frontend
• full-stack
• html
• html5
• javascript
• knn
• machine-learning
• natural-language-processing
• nltk
• pdfplumber
• python
• tf-idf
• vuejs