Eugene Yan's latest post proposes a low-effort practice for knowledge workers, including those in AI and ML: a weekly '15-5'—15 minutes to write five bullet points about what you did. The idea is to make documentation a regular, lightweight habit rather than a burdensome task.
According to Yan, this routine serves two main purposes. It makes your work visible to teammates and stakeholders, and it helps build trust over time. In fast-moving fields like machine learning, where much of the work is experimental and easy to overlook, a short weekly record can keep your contributions on the radar.
The post does not present new technical research or tools; it is a productivity and career-advice piece. For AI engineers who often focus on models and metrics, the suggestion is to also invest a small amount of time in communication. The barrier to entry is low: 15 minutes, once a week, with no need for polished prose.