From friend recommendations and newspaper advertisements to job search engines such as Indeed.com, the recruitment industry has been continuously evolving. LinkedIn has now taken that experience to a whole new level.

LinkedIn knows where you should apply, whom you should connect with and how your qualifications stack up against other applicants as you search for your dream job.

How does Linkedin do this? Machine Learning.

According to Anuj Goyal, a data scientist who works at Linkedin, Machine Learning forms the core of their recommendation engine.

Anuj has worked on several cool projects that apply Machine Learning (more on that later!) alongside pursuing Computer Science in school and online. Now Anuj has (what Harvard Business Review calls) the “Sexiest Job of the 21st Century.”

We talked to him about what got him started with Machine Learning.

What were the difficulties you encountered while learning Machine Learning? How did you overcome them?

The problem I faced while taking theoretical courses like Machine Learning in school was concentrating for the entire duration of a class. The material is difficult and, at that time, I wished for shorter classes so that I could revise what was taught and solve problems related to the topic.

When I look at online courses teaching these topics, I realize online education has solved this problem by allowing students to learn at their own pace.

What is an interesting application of Machine Learning?

During school, I was intrigued by the problem of Information Overload in Social Media.

For example, on Twitter, users tend to follow more and more people over time. As a result, they receive an overwhelming number of tweets on their newsfeed, to the extent that it is sometimes troublesome to go through every new tweet they receive. In a research project, we implemented a ML algorithm to sort all the tweets a user has received within a day in the decreasing order of interestingness. The algorithm used many social and personal features to solve this problem.

What do you work on now at LinkedIn?

I work on the Jobs Recommendation (i.e. Jobs You May be Interested In) System at LinkedIn. We use various Text Analysis and ML Algorithms to show relevant jobs to every LinkedIn member.

We parse the textual content in a member’s profile and extract features like skills, seniority, and industry. Similar features are extracted from the content on the Job listing. Furthermore, a Logistic Regression model is learned to rank relevant jobs for a given member using these features.

How do you know that the LinkedIn Job Recommendation System works better than simply going to a career fair?

In a career fair, a job seeker visits different companies to find out if they have relevant positions. On the other hand, job recommendation systems turns this problem around.

Job recommendations can be seen as companies approaching a member whenever they have an appropriate job for them. This saves lot of time and effort on the job seeker’s side and they get matched to more appropriate positions online (from the comfort of their home) than having to go through a career fair.

What are the other projects at Linkedin that use Machine Learning?

ML plays an inherent role in almost everything you see on LinkedIn. All recommendations (Job, News, Company, Groups), Friend Suggestions (People You May Know), Social Feed Personalization and Personalized Search use Machine Learning on a large scale.

Do you have any suggestions for our Machine Learning students who might want to become a data scientist?

I’d advise students to work on a project side by side while studying ML. Projects make ML a very interesting topic and also help students in gaining appropriate experience. The experience students gain through working on a project will help them the most in their career as Data Scientist.

Do you think ML will help us survive a robot uprising?

ML *is* the only way to learn how to trick a robot!

Interested in machine learning? Build your skill set with us. Udacity’s three part Machine Learning course series, created with Georgia Tech for the Online Masters in Computer Science degree program includes courses on Supervised Learning, Unsupervised Learning, and Reinforcement Learning.

stat Learning

Siya Raj Purohit
Siya Raj Purohit
Siya is a Program Manager at Udacity where she helps build and launch Nanodegrees. She is very passionate about all things STEM and is the author of "Engineering America". She loves learning about new startups, trying out all apps and exploring the San Francisco Bay Area.