This quick piece is about leading, motivating, and coaching data science teams. What I have below is a combination of what worked and what I wished I had done and some of what I intend to do in the future. Since our time for reading long articles has dwindled considerably in recent times, and you are probably reading this on your way to work or worse…at work, I will keep it short and sweet and stick to bullet items that we are so used to seeing.
My Top 10 intangibles for Coach, motivate, and Retain
- Always hire for attitude, not skills, skills can be taught. This is a cliché, but it seems to work. One bad apple can ruin the team morale. Be ruthless in weeding them out if you can.
- Focus on their development, not yours, as they grow-you grow.
- Encourage them to attend conferences, meetups, external and internal events. You need to deal with your organizational bureaucracy so that they have the approvals and budgets in place to do so. Let them shine.
- Give them an opportunity to present the work they did (under your guidance as appropriate). It shows that you trust them and it can be a game changer.
- Pay them on par or more than the market (sometimes this is not in your control). This market is very competitive. People will leave if they know your competition pays them twice as much.
- Insulate them from the bureaucratic mess and internal politics. You don’t want your team worrying about squabbles that large organizations typically have.
- Do bi-weekly review sessions with each of them…take them out on a walk or a cup of coffee and connect with them at a personal level. There’s nothing worse than a boss who cares only about what you can do for them. And please don’t ask the same old..how’s the family question. Get to know them better.
- Set the right tone for the team. Positive. Positive. Positive.
- Your team is looking up to you for directions, ideas, and you better step up your game technically and otherwise.
- Quoting from one of IBMs core principles, there are always ‘moments of impact.’ These are moments when a team member may have a personal challenge and how you react is something they will carry with them for the rest of the their lives..so be nice, seriously nice. Put yourself in their shoes and see how you may want your manager to react and do exactly that.
My Top 10 from the nerdy side of things for data science teams:
- Challenge them constantly on the way the models have been developed.
- Lead review sessions where ideas are presented and insist that they need to be taken seriously with rigorous prep work before they attend.
- If you have the ability, get into the nitty-gritty’s of things. Else, get help and have them review along with you. Dive deep into the mathematical and statistical details. If you are an experienced data science / analytics person, this should be straightforward for you. Ban power-point slides and force your team to go to the whiteboard.
- Be ready to conduct trainings yourself. Nothing demonstrates leadership better than doing it your self.
- Be ready to roll up your sleeves and write that R / Python / SQL code. This will earn you respect and trust.
- Encourage them to create new R packages. This helps in reducing redoing elements of work across engagements.
- Encourage / require them to contribute to stackoverflow, not just seek help from this great resource.
- Encourage them to participate on Kaggle competitions
- Encourage the senior members of your team to publish technical articles
- Encourage patenting, it encourages application-oriented thinking.
Doing all of the above I think will serve you well. I encourage you to comment below. I want to make this a top100 list of things that work for Data Science leaders.