How It Works with general skills as input
How It Works with a LinkedIn URL as input

SBUHacks2022

Stony Brook University Hackathon Repo

Team Members:

Paulina Chang, Mandy Tang, Jerry Zhu, Ethan Taqbachneck

UHired:

A linkedin-url based optimization bot, built to combat the possible bias that comes from an ATS Scan during the job application process, through the usage of NLP.

What is an ATS?

ATS stands for Applicant Tracking System, and is used by mainly software recruiters and employers to track candidates through the recruiting and hiring process. Because an ATS is used to further simplify and speed up the hiring process, candidates who don't pass the filter will automatically be rejected.

With UHired, you can receive valuable feedback/data to help you pass those filters and secure your dream job.

How It Works:

1) User pastes linkedin-url for the job they are applying for in UHired's search bar. 2) Outputs meaningful and coherent sentences, generated specifically by webscraping the key elements the ATS software looks for, using Natural Language Processing (NLP).

UHired is able to webscrape any arbitrary Linkedin-URL page (plus error handling if anything goes wrong) by targeting the structure of the page (lists) and extracting those relevant structures. UHired's general interface is powered by co:here's NLP system, with the model trained on various technical job listings. This allows for the generation of believable information that the user can use to better express their own skills on their resume or Linkedin.

3) Now, users can properly gauge what recruiters and hiring managers are looking for in their resumes and Linkedin profiles. What the users decide to do with the data is completely up to them.

Next Steps:

Not limiting the user input to just Linkedin URLs (ex: the user might be able to input a role at a specific company)

Resources:

https://github.com/cohere-ai/notebooks/blob/main/notebooks/Entity_Extrcation_with_Generative_Language_Models.ipynb https://docs.cohere.ai/entity-extraction

Link to Our Presentation: https://docs.google.com/presentation/d/196bN3w8RYdARJl0QmqAbUOwFhJOP7gLQ4OrCbSMWDIE/edit?usp=sharing

Built With

Submitted to

SBUHacks

Created by

Worked mostly on the back end. Specifically wrote the code for the web scraper, worked with Co:here to model, wrote the process for parser that found the job qualifications on the listings, as well as helped to set up the connection between the front end and back end.

Ethan Tabachneck
Front end mostly, otherwise some consulting and helping for other team members basically. Primarily responsible for the generative text feature. Done by lovingly forcefeeding a few listings into the model.

Jerry Zhu
Paulina Chang
Mandy Tang

Updates

Ethan Tabachneck started this project — Sep 25, 2022 06:47 AM EDT

Leave feedback in the comments!

Log in or sign up for Devpost to join the conversation.