Unpopular Opinion: Data Scientists Should be More End-to-End

Eugene Yan · 5 years ago

An opinion article argues that data scientists should work end-to-end across problem identification, data engineering, modeling, and deployment rather than specializing in narrow roles, citing reduced coordination overhead and faster iteration. The author references examples from companies like Stitch Fix and Netflix and notes that end-to-end ownership increases context, accountability, and job satisfaction through autonomy and purpose. However, the article acknowledges that specialization remains valuable for machine learning research, algorithmic trading, and other high-leverage domains where deep expertise is essential.