Why I Left the Data Science Realm(rachekintech.com)
rachekintech.com
Why I Left the Data Science Realm
https://rachekintech.com/2022/04/19/why-i-left-the-data-science-realm/
1 comments
Agreed! I think all new/aspiring DS should be told that even if we give a technically great argument, the PM/decision-maker might not adopt it. This surprised me & my colleagues when we first learned it.
I've heard of different DS organizational models, whether embedded in a cross-functional team like this DS, or on a DS-only team. The main downside I've heard for a DS-only team is that stakeholders just come to you for requests, and you don't build up domain expertise.
For embedded DS, how could we get them to have more decision-making power? The only solution I can think of for a product team is to have the PM have DS skills, which eliminates the need to have a separate DS on the team in the first place!
I've heard of different DS organizational models, whether embedded in a cross-functional team like this DS, or on a DS-only team. The main downside I've heard for a DS-only team is that stakeholders just come to you for requests, and you don't build up domain expertise.
For embedded DS, how could we get them to have more decision-making power? The only solution I can think of for a product team is to have the PM have DS skills, which eliminates the need to have a separate DS on the team in the first place!
> I think of product data scientists as consultants to the product. They are expected to have a deep understanding of how users engage with the product, and use these data insights to actively suggest improvements. Since their job is to just focus on the data and metrics, they are the best equipped to do this because they have spent the most time understanding what’s actually going on.
> However, the final decision on whether those suggested improvements are prioritized is made by the PM and engineering counterparts.
The first two points, about always being able to advance yourself, about social ordering, are also good framings to evaluate jobs from. Something about the third point feels more like an angle I'd want to stress so heavily, of where a role places us: interior or exterior to the thing, the work. There's for sure a lot of good in specialist & support roles, but recognizing the distance from the core that that sometimes has is key.
Different orgs ought to be able to explore different configurations. Data Science doesnt feel like it should necessarily always be this way! But there's definitely a gap, it would entail brdging responsibilitiea out away from a core group, or embeddding more people into a core, & that re-distributing ownership woupd require some organizational pioneering.