Software Developers for AI Solutions, but only with Data Scientists
Hire Software Developers for AI Solutions, but you still need Data Scientists. I see posts lately of folks not needing data scientists for their AI solutions, which is a recipe for failure.
By all means, hire software developers to build AI solutions. But these software developers always seek out the guidance of data scientists, as they know that anything that follows the data science lifecycle will break down and degrade over time, needing expertise at handling whatever went wrong within that lifecycle. You can’t count out data scientists- those software developers always seek us out- believe me, I know this fact first-hand, as I get pulled in all the time.
Don’t trivialize the importance of data scientists please. I see all of your posts on this, and it’s just not true-y’all need data scientists for AI solutions.
I think with AI solutions today, the breakdown of "who does what" has gotten complicated. Traditionally, DS focused on model builds, monitoring, and analysis. In the GenAI space, model building at the enterprise level is probably not as prevalent. Instead, adapting models is the focus.
I think a lot of teams are struggling with clear delineation of roles/responsibilities. DS could do a lot of model adaptations and monitoring, but I think SWE could as well.
I'd be curious to hear or see what other teams have found helpful in terms of each role's focus and how you structure your teams' and work to accommmodate.