Data science work often ends in charts, dashboards, PDFs, slides, or web reports. These outputs may be reproducible, beautiful, and technically correct, while still being hard or impossible for many people to use.

This talk looks at the accessibility gap between analysis and publication: color choices, typography, chart structure, document semantics, and generated PDFs. We will see why and how accessibility can become part of the data workflow itself, instead of a manual cleanup step at the end, or even not a step at all.