Note: See Andi's original LinkedIn post on the risks of building AI products with developer mentality, which got a shout out by the famous DJ Patil, former White House Chief Data Scientist: http://bit.ly/2uOhOou.
I have seen many data science teams dismissed after 6-12 months from both, large corporations & startups. One of the key issues is that managers treat data scientists as developers, and in turn data scientists take on the role as developers.
We know how to evaluate developers: lines & quality of code written, number of bugs tickets closed.
In data science, it is important to formulate problems well & align it with the business problem looking to solve, put out prototypes quickly, design experiments from a business idea, and to interpret the results with key stakeholders.
Managers also need to understand that data science is more R&D than software engineering. It is an investment that should from early on be clearly aligned with the business problem, data sources, and very importantly, team capabilities.
That is why, at Byteflows, we have teams of researchers with experience is exact sciences who understand the value of scientific discovery. It is the process of asking questions, letting the data take you to new places. To the unknown. To scientists, the unknown is part of the process. We are not afraid of it, we welcome it even though we may fail. Failure in part of the process. Very often we can see the light at the end of the tunnel and we don't care much about anything till we get there.
Kommentare