6 Lessons From 6 Years of Building Data Science Teams

Building Data Science teams is a challenge that goes beyond the hiring process. As I built out my Strategies for Hiring Data Scientists class, I had a chance to pull together my experiences from the last 6 years. 6 themes came up repeatedly.

Vin Vashishta | Originally Published: May 20th, 2021

Book time with me for Career Coaching or sign up for my Business Strategy for Data Scientists class.

Clients bring me in to build out their AI Strategy and Product Roadmap. That usually leads to ramping up their capabilities and expanding their Machine Learning teams. I didn’t think of the recruiting side of my work until a recruiter asked me to teach them how I built out Machine Learning teams so quickly. It kind of hit me, I have inadvertently added Technical Recruiter to the list of hats I wear for clients.

Here are 6 tips for building out Data Science teams from a Data Scientist who plays a recruiter from time to time.

Resumes and Job Descriptions Aren’t As Important As You Think

There is no way to really to explain what a person will do on a daily basis in a job description. The best you can do is cover the most important points.

  • How will that person create value?
  • What capabilities do most successful people in that role possess?
  • What people will the person interact with?
  • What work products do the produce and outcomes do they achieve?

  • The same is true of a resume. People are too complicated to be evaluated or screened with a pdf. You end up spending a lot of time interviewing if that is your only evaluation point. Looking at someone’s body of work is a better view of their capabilities and prior work.

    Relationships and Referrals Work Better Than An ATS

    I helped build an ATS and I honestly am not a fan of them for knowledge workers. I create a pipeline of talent by spending the time building relationships with Data Scientists. A few hours of community building a week pays dividends when it comes time to staff up a new team.

    Community building helps me understand a candidate better; back to resumes don’t matter much. I get a higher response rate when it’s time to hire because there’s a relationship. Communities are self-sustaining so I do not run out of talented people and I’m not actively trying to drag people into my network.

    Know What You Expect The Person To Do At A Project Level

    I get great feedback from people I hire around this more than any other part of my process. People like the certainty I have around what they’ll be doing. I talk project specifics and have a solid plan for the year.

    Once I’ve built out the Product Roadmap, I line up new head count and projects. From the initial phone screen, my questions are connected to finding out if they can contribute to the project in the way the business needs them to. When they ask about the role, I can answer everything from tech stack to work products.

    Keep Interviews As Minimalist As Possible

    I know what I am looking for; back to my last point. The interview process is standardized for each role. It is streamlined so it takes as little team bandwidth as possible, and I am always looking to slim the process down.

    I have a phone screen and a single team interview. Rarely, I need a follow up, but I cap it at 2 rounds. Interviews should be no more than 2 hours. Seriously, what are you going to learn in a 3rd round or hour that you haven’t already figured out? At that point it becomes an endurance race more than a capabilities assessment.

    Get All Interviewers On The Same Page

    Consistent questions, answer evaluation criteria, and interview flow. I talk to everyone for 5 minutes pr-interview to give them the background from my phone screen. I make sure everyone knows their role as an evaluator. We go in as a team and the goal is to figure out if the person is capable of doing the job.

    I keep the interview to the script and reign in questions that get too far out into the weeds. I remind everyone of the time box. Most importantly, I keep the tone professional and open.

    Interview To Include Not Exclude

    I emphasize that the candidate must be capable of doing the job, not the perfect candidate. We will be searching forever if the goal is “the best.” Let’s be real, all of us could be better. No one is fully baked if they have a growth mindset so capable will quickly become high performing on a supportive team.

    I want to fill the chair with the first person who can return the expected value to the business. That is the evaluation criteria. I hammer this home with the hiring team. “Are they capable of doing the day to day, not the one off or extraordinary asks?” If the answer’s yes, let’s send them an offer and get back to doing our jobs. We’re Data Scientists, not recruiters, even though I sometimes play one on TV.