Eugene Yan's mailbag post addresses two recurring questions at the intersection of data science and business: whether DS work has business requirements, and when it makes sense to split DS and DE roles. The answer to the first is a clear yes. Data science is not a purely technical exercise; it exists to serve business goals, so requirements like success metrics, stakeholder alignment, and deployment constraints are part of the job.
On the second question, Yan argues that the split between DS and DE is not a binary choice but a function of context. In small teams, a single person may need to handle both modeling and data plumbing. As the team grows, the volume and complexity of data engineering tasks can justify a dedicated role, allowing each function to focus on its core strengths.
The key takeaway is that there is no universal rule. Organizations should evaluate their own bandwidth, project pipeline, and technical debt when deciding whether to separate these disciplines.