Kaggle Career Con 2018: Day 2 Recap

Day two of Kaggle Career Con 2018 commenced today at noon PT with a series of livestreams available on Kaggle's YouTube Channel, focusing on data science resumes and the interview process.


Session #1 Recap: Portfolio and Resume Analysis with Data Science Hiring Managers

This session consists of a panel of 3 data scientists from Quora, Kaggle and Two Sigma giving their opinions on several resumes submitted by Career Con attendees. The advice and opinions given in this session were mostly specific to the resumes that they happened to review and the panelists didn't agree on how best to present information on resumes. In addition, the resumes reviewed all featured professionals with several years of experience and often advanced degrees--background which is likely beyond most Kaggle Con attendees. The lack of agreement between panelists meant this session was not particularly constructive and worse, the general sentiment among users in the YouTube chat and slack suggested this session was disheartening to many attendees. This is one to skip.


Session #2 Recap: Overview of the Data Science Interview Process

The second session of the day featured two engineers from Google, discussing that data science interview process and about how to approach different aspects of interviewing. A few of the key points raised included:

1. A data science interview measures your ability to interview for a data science position, not necessarily to actually do the job, although the two should be correlated.
2. Find out as much as you can about the company you are interviewing with ahead of time.
4. Don't overstate your skills on your resume.
5. Be honest about what you don't know; be intellectually humble.
6. There's a good deal of arbitrariness/randomness in the process--don't dwell too much rejections: move on to the next opportunity.
7. Be your authentic self.
8. Interviews are not a one way street. You are also there to figure out of the company is right for you. Make sure a company's core values align with your own and have questions to ask.
9. Mock interviews are useful.
10. Be able to talk about specific experiences to show your enthusiasm and what you care about.

Although the interview process at Google probably differs from a lot of other companies, the session was very positive and contained actionable advice that should be useful in any interview setting.


Session #3 Recap: Live Breakdown of Common Data Science Interview Questions

The final session of the day consisted of a mock interview with a live coding component presented by data scientists at Kaggle. The session covered a variety of considerations for both preparing for an interview in advance and for performing well during and interview. There was only one mock coding question in which the candidate was asked to create a function to identify outliers in a column of data. Outliers should show themselves naturally during the EDA process--a boxplot is enough to check for outliers--and there is no fixed definition as to what constitutes an outlier so the problem itself was flawed, but that aside, it was still useful as an example of how to plan, think and talk your way through a problem. Of all the sessions thus far this was the most geared toward people looking for a first data science job who might be coming straight out of college or from a field with little to no overlap with data science. It was a stark contrast with session #1 of the day, where panelists acted like resumes submitted by experienced professionals in closely related fields and PHDs were barely suitable for junior data science jobs.