- What skills do you want the Data Scientist to have?
- What do you want to know about the Data Scientist?
- What questions can you ask to answer the above questions?
- How did the candidate contribute to previous projects.
- What tools and skills does the candidate use the most and how do they use them to achieve their goals.
- How does the candidate think about new and/or unfamiliar data.
- How does the candidate respond to being challenged and/or stressed.
- How will the candidate fit in the current team dynamic.
A candidate must pass all parts and exceed in at least two or three to be considered being hired. This might sound a bit draconian, but each of the five parts are integral to the hiring process. Here are my overall thoughts when running through the interview on a part by part basis:
- If a candidate has been a team manager for many years with little recent hands on experience, it brings into question how well they would do as an individual contributor.
- What are the goto tools for the person, e.g. how do they attack a problem?
- Data Scientists are exposed to new situations every day and must be quick on the understanding and integration of this data into analysis or systems.
- Anybody will come to a point where their ideas or techniques will be challenged in an adversarial way. It is good to get a feeling how someone handles this type of situation.
- Team dynamics is one the most important pieces of a company. A candidate must get along with the interviewers, else they will most likely cause strife within the company (This is most important in a small startup environment where everybody interacts with everybody else.)
Most questions I ask cover at least two of the five points, and I have found the best flow for interviewing follows a 1 to 3 order. While question the candidate, I include points four and five when relevant.
- Programing and algorithms
- Data Modeling
- Statistics and Methodology (e.g. frequentest vs Bayesian approaches)
- System architecture, data flows, and technology
- Interpersonal skills
Of course, the questions I ask come from not only the level and experience of the candidate, but the flow of the interview and what seems appropriate at the time.
Andrew Eichenbaum is VP, Data Science Solutions at DesignMind. He specializes in data mining, data modeling, and artificial intelligence. Andrew heads DesignMind’s Data Science division.