Good Data Scientist/Bad Data Scientist

The following post is inspired by "Good Product Manager/Bad Product Manager," a popular piece of tech writing penned by the venture capital firm Andreessen Horowitz. At Uptake, data science is in our DNA. In this post we share our perspectives on the qualities of good data scientists. Think you have what it takes to be an Uptake data scientist? Check out our open roles here.

Good Data Scientists know the market, the product, the product line, and the competition extremely well and operate from a strong basis of knowledge and confidence. Bad Data Scientists don’t think about the product. They just know what tasks they were given, but not how they will be ultimately surfaced to our customers to create value.

Good Data Scientists measure the value they are creating for their customers, in terms the customer understands. Bad Data Scientists just present statistical metrics without any context or measurement of business value for the customer.

Good Data Scientists value effectiveness over complexity. Good Data Scientists start with simple solutions and increase the complexity, while comparing the additional value added versus the additional complexity. Bad Data Scientists implement trendy solutions. Bad Data Scientists value what’s cutting-edge or cool, versus what will deliver the most value for their customers. Bad Data Scientists feel best about themselves when they have done something complicated and bemoan "incompetent" project managers (PMs), Engineers, Sales People and other Data Scientists who don't understand their elegant work.

Good Data Scientists are able to align their project deliverables with product, engineering or other stakeholders. Bad Data Scientists do not align their projects with product, engineering, or other stakeholders. Good Data Scientists don't waste time working on projects with poorly defined endpoints or handoffs to other teams. Good Data Scientists recognize these time-wasters and proactively create and write down agreements to facilitate organization across teams. Good Data Scientists understand a model isn't valuable until it's in a product and vetted by users. Bad Data Scientists think that getting the best statistical performance is the most important objective.

Good Data Scientists reuse a lot of code. They have a strong knowledge of the existing tools at their disposal and frequently leverage them. When Good Data Scientists discover a gap in an existing tool, they work to fill that gap in a way such that others can later leverage their work. If that cannot be done at the time, Good Data Scientists log that feature request in the (Asana) backlog for future prioritization. Good Data Scientists make sure their work is repeatable and documented because they expect others will use their work. Bad Data Scientists write a lot of code. They routinely open up their text editor to a blank page and start coding. When Bad Data Scientists discover a gap in an existing toolset, they develop ad-hoc work arounds without logging the feature request or thinking about future modelers. Bad Data Scientists like to be the only source of knowledge for a particular domain.

Good Data Scientists update their (Asana) task lists and project status every week, because they value discipline. Good Data Scientists then do everything in their power to hit those deadlines. Bad Data Scientists do not update their (Asana) task lists and project status every week, because they do not value discipline. Bad Data Scientists complain they are overworked or that they didn’t properly estimate how long something could take.

Good Data Scientists are curious and are not afraid to ask questions when they don’t understand something. Bad Data Scientists are afraid they might sound stupid if they ask something that they think they should already know.