Lately, a number of my clients have asked me to help them scope and recruit for the role of chief data scientist (a.k.a. ‘Head of Data Science’, ‘Lead Data Scientist’, ‘Head of Analytics’, etc.).
The roles are typically opened by the client company for one of two reasons:
- the company is looking to launch a new data science department, or
- the company has already tried to launch a data science department leveraging existing management and has realized the hard way the need to bring in fresh, specialized leadership.
Over the past 20 years, I’ve probably conducted several hundred interviews for analytic roles (after screening many, many more C.V.s). The candidates with whom I’ve spoken have come from all over the world, many having completed world-class technical graduate programs or MBAs from schools such as Wharton, Chicago Booth or Oxford. It’s been a real privilege to find and hire many excellent people over the years.
Filling a Chief Data Scientist role, however, is particularly challenging because of the complex requirements which the successful candidate must satisfy:
Possession of three unrelated skill sets
The chief data scientist role requires a unique blend of technical, business and communication skills; skills which often correlate negatively with one other. That is to say, individuals who excel in technical areas often have proportionately less interest in mastering the art of communication with non-technical business colleagues. They often do not want to confine their efforts to tasks with demonstrable business value. The successful chief data scientist, however, will not succeed unless he or she possesses both strong communication skills and strong business acumen.
Breadth and depth of technical skills
From an analytic perspective, the role requires both familiarity with a broad range of tools and techniques and also an experience-based understanding of what is involved with in-depth technical implementations. There is certainly space in an organization for specialists in areas such as statistics, deep learning, NLP, or integer programming, but for the lead role, the right candidate must have an overview of the entire analytic tool chest, so that he or she can choose analytic techniques that best address the business problems at hand.
Even more, the candidate must be familiar with analytic tooling, including data base technologies (e.g. SQL, no-SQL, graph databases), programming frameworks (e.g. Spark, MapReduce), development languages (python, R, Matlab, Java), and prototyping tools (Rapidminer, Enterprise Miner, Knime).
The technology space is already quite broad and is developing very rapidly. Properly leveraging existing technologies can easily save months or years of in-house development.
Ability to deliver results
I love Joel Spolsky’s guerilla guide to interviewing. At the risk of oversimplifying, Joel looks for only two things in a candidate: 1) intelligence and 2) the ability to get things done. In the case of a chief data scientist, initiatives will almost certainly not ‘get done’ if the leader is not able to
- Understand tangible business drivers and KPI’s
- Identify appropriate data science techniques, tools, and applications, typically drawn from cross-industry studies
- Communicate vision and win buy-in from peers
- Execute the analytics projects in a lean manner
I’ve spoken to quite a number of smart candidates with great backgrounds but without a track-record that convinces me they will deliver amazing results in a business environment. I always pass on these candidates for lead roles.
Three phases of the hiring process
There are three main phases through which I typically progress alongside the recruiting client company
- Working with the recruitment team
My favorite part of the process. The men and women in recruitment are always a lot of fun to work with and very enthusiastic in learning how to recruit new profiles. The chief data scientist role is almost always completely new to them in terms of skill sets, technologies, background and business experience, and so I find it necessarily to work very closely with the internal recruiter(s) over multiple sessions in order to scope the role properly, to identify appropriate distribution channels, and to sketch the profiles of candidates to whom they should reach out.
I also work with both recruitment and with the hiring executive to develop a salary indication. It’s important to think about salary early, as many companies do not realize the high premium that this role commands in the job market these days. A company will lose qualified candidates if it takes them too long to bring their salary expectations up to a market-conforming level.
I find it helpful to continue working very closely with the recruiter and the hiring executive throughout the hiring process.
- Finding strong candidates
The most challenging part. We are looking for an individual who can take complete ownership of the analytic program within a company. I described some of the inherent difficulties above. The techniques for finding and screening the candidates would require another, very long article.
- Landing the candidate
The top candidates for this role will have many job options. Offer a competitive salary and follow up closely with the candidate to quickly address any ancillary concerns. As with most positions, geography often plays an important role in a candidate’s decision, so the clever recruiter will keep this in mind in targeting candidates.
For lead data science roles, my experience is that the really strong candidates will be drawn most by the opportunity to work with interesting and abundant data and by the opportunity to contribute to growing an interesting business in creative and meaningful ways. These top candidates are people who love applying complex analytics to large data sets using innovative tooling in order to bring tangible results to real-world business problems. In essence, these are the candidates who want nothing more than to bring real growth to your business.