2. We have a problem
● 60% say the roles are not
clearly defined
● 50% found the recruiting
process to be ill-defined
Source: CutShort Data Science Interview Survey 2018
The result - you can’t hire. Or worse, make a bad hire!
4. Recruiters
● Job titles are vague and very broad
● Lack of clear functional responsibilities
● Business has unrealistic expectations from a single role
5. What recruiters end up doing
● Search on keywords.
○ Bad idea
● Filter on top institutes
○ Limits options
● Filter on related companies
○ Every company has different data maturity
○ Titles and job functions may not match
And yeah, a lot of prayers
8. Sit with your team and
● Remove the job title from your mind
● Understand the business problem
● What will each role do?
○ Work with huge amounts of data?
○ Slice and dice data to draw historic insights and
reports?
○ Find patterns in data to predict something
○ Something else?
9. Step 1: Map the role to skill areas
● 4 key skills areas
○ Maths and Stats
○ Business skills
○ Data proficiency
○ Programming
● Each is very different
○ So choose maximum 2 areas
11. Step 3: Now find the job title
Core skill areas Skills/Traits Job title
Business skills + Stats Excel, Tableau, Qlikview Business Analyst, Data Analyst,
Data visualization expert,
Business Skills + Maths & Stats Mathematical models, regression,
CNN, Random Forest,
Data Scientist
Programming + Maths & Stats Using and enhancing libraries
Tensorflow, Python, Java
ML Engineer
Programming + Data proficiency Databases, algorithms, BigData,
Java, Spark
Data Engineer, Big Data Engineer,
Software Engineer (Data)
12. Now, recruit!
● Choose the right place
● Specific targets
○ For business skills - look at blog posts
○ For ML and Data Scientist roles, look at Kaggle
& Stackoverflow activity
14. How to screen
● Define primary and second skill areas clearly
● If you’re not a techie, you have an edge!
Example screening conversation:
You: So what do you do?
Candidate: We build models to decide optimal driver locations.
You: Cool. How?
Candidate: We use Bayesian neural network
You: Aah. How would you explain that to a layman like me?
15. Summing it up
● Do your homework
○ Choose max 2 core skills
○ Make 1 primary, 1 secondary
● Your context is important. Stop looking at competition.
● Hiring in India
○ Limited talent: Look beyond premium institutes. Example UpGrad talent pool
○ Second skill areas can be built. Look for fundamentals and attitude.
○ Reduce reliance on keywords - look for competencies