Le Nguyen The Dat bio photo

Le Nguyen The Dat

Data Science and Engineering at Uber. Kaggle Master.

Email Twitter Facebook LinkedIn Github

image

What are they, where do they fit in?

The data team is probably the most impactful team in your entire company. Their products either directly help the key management people to make critical business decisions, or simply generate more revenue by giving insights and actionable recommendations with statistically proven approaches.

The data team sits in the center of your company - they don’t hide away in their own island, they don’t do work that is not relevant to any part of the company. Their direct end-users and stakeholders are the people in various functions of the company.

What skills do they need?

A successful data team needs deep technical knowledge. Especially in term of engineering and data:

  • They decide how the company’s data warehouse / data mart distributed infrastructure looks like, either it’s Hadoop, Spark, RDBMS, NoSQL, or any other hybrid solution that’s available on the market.
  • They build data products, whether it’s ETL, reporting, or data science related.
  • They have a practical and business mindset, they should be able to utilize, to leverage the enormous amount of data they collect, clean, and control everyday.

Not only that, they are in charge of the whole company data structures, the single source of truth. They need to understand every single piece of the data inside every single function of the company.

How do they work?

The data team needs to value transparency, which can be achieved easily through: Github issues, JIRA tickets, Trello, or any kind of project management platforms:

  • They needs to talk to and work with their stakeholders.

Think of it as working in a restaurant or simply a fruit juice stand, everyone else in the company are your customers - you need to ask them what they want, how they want it, and bake your products accordingly. Don’t just blindly go work on something no one needs, understands or even cares about, that is just wasting the company resources.

  • Deliverables should be produced, progress should be clear.

It’s not rocket science, with data, there are tons of things you can do to help your company. The data team need to be able to build prototypes, test hypotheses, or basically fix data-related problems on a weekly basis. If your current data team has spent a month or two without any single deliverable, not even documentation or any short of work, progress you can see or understand, you might want to audit their work, or might as well just fire them all and get a better team.

Lastly

Data team is not pure development team. They should know how to build system but not keeping building and trying new system everyday. It’s simply not effective and not fun either. Stop re-inventing the wheel, stick to one system that works, improve it, build more data products, cool reports, effective statistical models.

And that, to me, is the real data team.