Data scientist or ML engineer? Scope the role before the search

A large share of failed data hires were doomed at the JD stage: the company hired a data scientist for an engineering problem or an ML engineer for an insight problem. The titles blur together in job posts, but the day jobs, and the people who thrive in them, are distinct.

What each role actually does

A data scientist turns questions into analysis: experiments, models as investigative tools, insight that changes decisions. Their output is understanding. An ML engineer turns models into software: pipelines, serving infrastructure, latency, monitoring, retraining. Their output is a system that works at three in the morning. Analytics engineers, a third species, build the data foundations both depend on.

The diagnostic questions

Ask what artifact you need in six months. If the answer is “we finally understand churn and price correctly,” that is a data scientist. If it is “the recommendation model runs in production and does not fall over,” that is an ML engineer. If it is “our dashboards stop lying,” you need analytics engineering before either. Then ask who reviews their work: no senior engineer around means your ML engineer needs production self-sufficiency; no product counterpart means your data scientist must self-direct toward business questions.

The startup reality of hybrid hires

Early teams often need fractions of each role and try to hire one person who does it all. Unicorns exist but are rare and expensive; more often you get someone strong in one mode and quietly drowning in the other. If you must hire hybrid, decide which mode dominates the first year and screen hard for that one, treating the second as a bonus.

Scope, then search

The search itself follows standard senior-hiring rules: direct sourcing, evidence of shipped work, and fit screening for stage and working style. But none of it saves a mis-scoped role.

Scoping is the first step of every search we run, and it is where data roles most need it. Unsure which job your problem is? Book a demo and we will scope it with you before anyone writes a JD.