SageMaker Data Prep Tool Recommendation Framework
AWS SageMaker
Recommendation framework
Product and user knowledge as the foundation for content judgment — the recommendation framework required understanding both the tools and who shows up to use them. The form was mandated. The content required knowing the product and the user.
The situation
SageMaker's data preparation surface had grown to include multiple overlapping tools — Data Wrangler, SQL integration in Studio, EMR, EMR Serverless, Glue — with no top-level guidance helping users understand which to use when. A guide-wide campaign was launched to create orientation nodes at the top of major developer guide sections. The data prep section was a real pain point: users were confused, and the tool sprawl was legitimate.
The task
Own the data prep recommendation framework — make the content decisions that would guide users to the right tool for their use case, within a mandated template structure.
What I did
Drew on three years of staying current with the SageMaker ecosystem — reading feature announcements, testing tools hands-on, building familiarity beyond the immediate scope of assigned work. Mapped three primary use cases to recommended tools, leaving significant options out deliberately — a framework that covers everything guides no one. The hard part was knowing the users well enough to make that call confidently. The form was given. The rest was earned.
What happened
A recommendation framework that gives users a genuine decision path across a fragmented tool surface. Arrived at PM review already correct — no brief to work from.
Product knowledge depthContent strategyEditorial curationInformation architecture