I decided to treat data‑annotation platforms and data‑warehouses as two sides of the same coin, creating a shared repository of annotated datasets that can be reused across teams.
I assert that 50%+ of the effort in an AI development project goes into data engineering, including annotation work.
In my experience, data quality drives AI outcomes, so I treat data engineers as the backbone of any ML initiative. This view aligns with my earlier notes on operationalizing data quality operationalizing data quality in practice.