In this talk, Dr Emanuele discusses one of the most intimidating and fateful parts of data science job searches: the technical interview. He discusses all the preparation aspiring and current data scientists should have as part of their routine, and reveals intimate insights behind how he interviews, vets, and hires data scientists in his startup.
Landing your first Data Science Job: The Technical Interview
1. Landing your first Data Science
Job: The Technical Interview
Vincent A. Emanuele II, Ph.D
vincent.emanuele@anidata.org
November 3, 2016
2. Technical Interview Preparation
You need luck to get a job
Luck = Preparation + Opportunity
Most of my talk is about good preparation
habits so that you have very little “extra” to
stress about before your technical interview
You can start applying this advice
IMMEDIATELY
The longer you apply this advice, the more
prepared you will be for a tech interview
3. Three keys to doing well in a data technical interview
● Know yourself and your purpose
● Make sure you know one thing in great detail
● Demonstrate that you keep up with the state of the art, even if you don’t really
understand it or know the technical details
5. Professional Bio
● PhD in Electrical and Computer Engineering from Georgia Tech (Signal
Processing and Machine Learning) (2010)
● CDC Visiting Scientist (2006 - 2013)
● First data scientist at Wellcentive. Founded Data Quality, Data Governance, and
Data Science Teams (2013 - 2016)
● Co-founder of Anidata (2016)
● Founder of Zylinium Research (2016)
6. What am I thinking when I interview you? What is in the back of my head?
7. My biggest worry for your on-site TECHNICAL interview
“People go into startups thinking that the technical problems are the challenges… No,
every real problem in startups is a people problem, and as such they’re the hardest to
solve, as they often don’t have a real solution… Startups are experiments in group
psychology.”
- A. Martinez in Chaos Monkeys: Obscene Fortune and Random Failure in Silicon
Valley
8. First thing I need to see very clearly: Your Purpose
Why are you here? Why are you
interviewing at this company?
Why is this important for me to
understand?
One word: HARDSHIP. Your answer to
this questions gives me insight into how
much grit you will have to push through
hardship. And you WILL encounter
hardship.
9. Bad reasons to work for me/red flags
I always considered myself a scientist
I just saw The Social Network and I heard startups are cool
I want to buy a BMW
What are some good reasons?
10. Types of Data Scientists. Know what you want!
Type A Data Scientist: The A is for Analysis. This type is primarily concerned with
making sense of data or working with it in a fairly static way. The Type A Data
Scientist is very similar to a statistician (and may be one) but knows all the practical
details of working with data that aren’t taught in the statistics curriculum: data
cleaning, methods for dealing with very large data sets, visualization, deep knowledge
of a particular domain, writing well about data, and so on.
https://medium.com/@rchang/my-two-year-journey-as-a-data-scientist-at-twitter-f0c132
98aee6#.8ufkrgg55
11. Types of Data Scientists. Know what you want!
Type B Data Scientist: The B is for Building. Type B Data Scientists share some
statistical background with Type A, but they are also very strong coders and may be
trained software engineers. The Type B Data Scientist is mainly interested in using
data “in production.” They build models which interact with users, often serving
recommendations (products, people you may know, ads, movies, search results).
https://medium.com/@rchang/my-two-year-journey-as-a-data-scientist-at-twitter-f0c132
98aee6#.8ufkrgg55
12. Clusters of Data Science Skillsets
Source: 2016 O’Reilly Data Science Salary Survey
13. Salary Increases and Tool/Skill Progression
Source: 2016 O’Reilly Data Science Salary Survey
14. Your goal on technical interview: Figure this out
What you want to do What the company
needs
15. How do you figure out what you want?
1. Talk to other data scientists and hear about their career experience
2. Get as much of your own experience as possible
3. Read blogs and books to learn about how people are doing data science elsewhere
16. How to figure out what companies are willing to pay for?
● A job listing is a statement: We are willing to pay you X for these Y skills.
● Be a data scientist, go collect your own data from LinkedIn and Indeed.com and
do some analysis
● Read data science salary surveys, but be careful and watch out for sample bias
19. Assessing Technical Mastery
I want you to tell me which “kick” you have practiced the most, and I want you to
show me. This is sufficient for understanding your ability to master the details.
Translation: I want you to choose what you say you know the best, and teach me about
it.
Implications for you: Make sure you know 1 thing on your resume in GREAT detail
Further, you need to SHOW me you mastered the details rather than TELL me. What’s
the difference?
20. Ways to SHOW mastery of details
Describe a failed project, all the pros/cons of design
considerations, and how you would do it different
Be able to derive important results on the whiteboard of
methodology used in your work
Know the most important publications on the topic you
worked on by First Author/Year, and be conversational
21. My technical evaluation red flags
Resume lists: “I am an expert in: <30 items>”
I used that technique because it’s state of the art
(with no further explanation)
I didn’t do that because it was too simple
I heard Google/Facebook are doing it
Being in love with data science vs being in love
with solving problems with data science
22. Does this person keep up with the state of the art?
Data Science evolves rapidly, but the fundamentals stay the same. Be prepared to
continuously learn the rest of your life.
Keeping up is important!
My recommended way: Read KDnuggets weekly newsletter.
You don’t need to know the technical details of the emerging trends, just understand
the basic idea of how people are trying to attack problems differently.
http://www.kdnuggets.com/
23. You are what you read
How many books published in 2016 will people still read in…
In 5 years
In 10 years
In 25 years
In 50 years
In 100 years?
24. You are what you read
The “classics” in a field influence all other works. There is 95% overlap in content, and
most “new” material is not very new or insightful.
In my research group we spend 50% of the time reading and rereading the “classics” in
machine learning, and the other 50% scanning for new papers.
You should think about knowing some of the “classic” papers in GREAT DETAIL.
That is a good investment of your time.
25. More about blogs and newsletters
The Data Science Geek equivalent of fashion trends and gossip (sometimes useful)
28. The three important take homes
● Know yourself and your purpose
● Make sure you know one thing in great detail
● Demonstrate that you keep up with the state of the art, even if you don’t really
understand it or know the technical details
29. Don’t wait.. Start preparing TODAY
Work on a project, or review a project that you’ve completed and learned in detail
Collect data on skills people are willing to pay for
Talk to other data scientists
Sign up for KDnuggets and read weekly
Start studying a “classic” machine learning paper
Decide which type of data scientist you want to become and what you are missing