

The Best Practices For Building Strong Data Science Teams
Career InsightsConferencesposted by Alex Amari May 16, 2018 Alex Amari

Key Points:
- Hiring managers and business leaders often overestimate data scientists’ capabilities, resulting in inefficiencies in the hiring process and poor performance from data science teams
- The pre-interview process is crucial. It should involve thoughtful contemplation about the role and why it’s needed, and effective phone screens to ensure that only qualified candidates make it to the interview stage.
- Hiring managers should strive to employ transparency and personability in trying to close a candidate. Good data scientists are likely to have a lot of offers, but if you can show them how they’ll fit in to work they’ll love with their specific strengths and skills, you’ll have a great chance of landing the hire.
In his talk at ODSC East 2018, Dr. Drew Conway, Founder and CEO at Alluvium, described the state of human capital in the world of data science and business and best practices for building a data science company. Across virtually all industries, businesses are beginning to see the potential of data and artificial intelligence to transform how they work. The result is what Conway calls a “Jurassic Park moment” of data science, where companies are working to build teams of ‘unicorn’ data scientists capable of solving their most difficult challenges. The problem is, many hiring managers and business leaders don’t know where to begin when it comes to attracting and retaining top data science talent for their organizations, much less about how to help new data scientists thrive at their companies. Using a series of examples from his own career, as well as hiring practices employed by Alluvium, Conway offers business leaders specific and practical advice to hire top data science teams.
Slide Copyright Drew Conway, ODSC East 2018
At Alluvium, the hiring process begins with the pre-interview process, where the challenge involves writing a specific job requirement and posting. Questions to ask at this stage include what will this new employee do and is it a realistic expectation? If a team of us can’t do that, then why are we hiring them in the first place? Next steps include building a good take-home that tests the skills directly involved in the position (possibly including the exact kinds of data and packages involved in the role), and conducting a phone screen to answer candidate questions and determine whether or not the person will succeed with the take-home.
Slide Copyright Drew Conway, ODSC East 2018
After the successful completion of a take-home begins interview process, including a code review and practical exercises to gauge the candidate’s interests and abilities in project discovery and planning. Here you’re looking for the candidate’s ability to see opportunities to benefit a client’s business, while also sketching out the details of a feasible project to make improvements happen. Of course, this is also a chance to see how the candidate interacts in a collaborative setting, and whether or not they’d be a good cultural fit.
Post-interview, Alluvium seeks team consensus on whether to hire the candidate or not. In the event of a consensus no-hire, you should ask where things didn’t go well from the pre-interview process to now. With a consensus hire, Conway emphasizes the value of transparency and personability when approaching the candidate with the offer. Top data scientists often have a lot of offers, but if you can really sell them on great work they’re going to do that they’ll absolutely love given their background, there’s a great chance they’ll choose you.
Check out Dr. Conway’s full talk below and learn more about how to build out a strong data science team in your organization.
ODSC East is not the end of 2018’s exploration of advancements in AI, machine learning and more. The future of AI continues to gather in San Francisco with ODSC West and in London with ODSC Europe. Be sure to connect to see who is involved in the upcoming talks, workshops and speaking sessions.
Stay up to date on all things data science here.