hireHER | Devpost

Inspiration

One of our teammates went to a conference and professionals said that one of the biggest issues in the tech industry is that women aren't applying to jobs. One of the reasons for this is because of the unconscious gender bias towards more masculine wording in job descriptions.

What it does

We created a responsive web app content editor that analyzes whether a job description contains subtle linguistic gender-coding*, resulting in discouragement when women are job seeking. A hiring manager can submit a job description and the application uses NLP machine learning to identify and highlight words that are statistically proven to be masculine. The user will then be shown a list of suggested gender-neutral replacement words that would reduce the bias within the description.

*Gender-coding is assigning exclusive traits exclusive to males or females.

How WE built it

We used CSS, Node.js, Javascript, and HTML to create the web application. Google Vision API was used to retrieve the text stored in Google Cloud Storage, to transform the pdf into the necessary format for the machine learning model. A natural language processing API was used to train a model to identify statistically proven masculine words used within the given job description.

Challenges we ran into

Most of us were new to javascript, so there were some difficulties and bugs created as we did not fully understand the syntax yet. We also had problems learning how to use git and training our model.

Accomplishments that we're proud of

Learning how to tokenize and classify words based on a model that we trained, and being able to utilize our current skills and explore further skills are things that we're proud of, which led us to finish our project within the 24 hours.

What we learned

We learned natural language processing, github, how to use node.js, and expanded our knowledge in Javascript.

What's next for hireHER

To be able to target other minorities, provide more suggestions for replacement words by training our model further, speed up the processing time, being able to take multiple paragraphs, and possibly create an extension to Grammarly to check grammar and spelling. We hope in the future to use Vision API to read images and other files types for greater flexibility.