Move from generalist 'data analyst' to analytics-engineer or ML-engineer in 12-18 months — own dbt + warehouse modeling end-to-end and ship one production pipeline, or take a Stanford-tier ML specialisation and apply it on real company data; band uplift is 30-60% and the role survives LLM commoditisation.

Ship one public technical artifact this year (a Kaggle medal, a public notebook on a real dataset, or a blog post series with code) — the network/visibility floor is the single biggest drag on transfer cost at this seniority and is cheaply fixable.

Aim for a name-brand or top-tier data-heavy employer (a hyperscaler, frontier-AI lab, or a recognised data-platform company) on the next move — at 26 the employer-brand jump compounds for the entire 35-year career ahead.