JobDisco

Inspiration

Personal and acquaintance's experiences and difficulties with finding the appropriate job/career.

What it does

It's a personalized recommendation engine that uses Machine Learning to suggest best matching jobs. It uses as input your skills, experiences, preferences and hobbies. LinkedIn's integration enables to quickly import the data and find recommendations.

How we built it

We built a web application with Node and React that communicates in the backend with a Python program that runs the recommendation engine itself.

Challenges we ran into

The biggest challenges have been:

• finding appropriate Machine Learning model for multi criteria recommendations
• finding extensive datasets that contain data anonymized career data (positions, skills, education)
• solving how to run a Python process within a Node web server
• connecting with LinkedIn API (OAuth client in JavaScript)

Accomplishments that we're proud of

We are happy that:

• we managed to finish a working prototype
• we spotted a real problem and our solution could be helpful for many people, including our friends

What we learned

Key lessons:

• the time constraints of a hackathon require to make some trade-offs between having something working and the scope of features it has
• even seemingly simple things like integrations require time and might slow down

What's next for JobDisco

Project roadmap:

• applying to Microsoft Partner program to get access to full Linkedin Profile API
• adapting the Machine Learning model to consider multiple criteria
• create an ML model to generate recommendations based on similar people
• searching for more extensive datasets with career information
• add an extention proposing to enhance a candidate's CV with recommended volunteer experience

Built With

• javascript
• machine-learning
• mdbootstrap
• node.js
• python
• react