2020 Retrospective: New Country, New Role, New Habit

Eugene Yan spent 2020 doing two hard things at once: starting a new job at Amazon and forcing himself to publish something every single week. He'd just finished a CS master's in 2019, moved from Singapore to Seattle in January, and landed on a lean recommender-systems team — one data scientist, two applied scientists, four engineers, all leaning on SageMaker and managed Spark to punch above their headcount. Then the pandemic hit, shelter-in-place kicked in, and the extra hours at home became fuel for a writing habit he'd previously kept at a monthly pace.

The numbers by year's end are the interesting part. Fifty-five posts, roughly 260,000 page views, and two Hacker News front-page moments that dwarfed everything else on his traffic graph — one about his path into data science, another about a Zettelkasten note-taking system. Everything else looked flat by comparison, though smaller bumps came from Twitter shares and mailing lists like O'Reilly Data & AI and Data Elixir. His email list, migrated from Substack to ConvertKit, climbed from 30 subscribers to a real audience, even as open rates drifted slightly downward over the year.

What actually resonated wasn't always what he expected. A post arguing data scientists should be more end-to-end — something he assumed would draw pushback for being too generalist — turned out to be one of his most shared pieces, with other practitioners piling on in agreement. A three-minute post called