MLE vs SWE vs Research Scientist

Behind the ML Engineer Title — Part 2: MLE vs SWE vs Research Scientist

I spent years confused about what these roles actually meant in practice.

Job descriptions don’t help. Everyone is “working on cutting edge ML systems” and “collaborating cross-functionally.” Cool. But what do you actually own? What do you actually do on a Tuesday?

Let me break it down the way I understand it from the inside.

The Software Engineer

An SWE in an ML org doesn’t touch models. They build the infrastructure that makes models possible — the serving layer, the data pipelines, the tooling that lets people like me run experiments without thinking about Kubernetes.

Critical work. Genuinely hard work. Just not ML work.

If you’re an SWE who wants to get closer to the models, you’re essentially looking at transitioning into an MLE role. The two jobs look similar on a job posting and feel completely different day to day.

The Machine Learning Engineer

This is my job. We build models, own models, and ship improvements.

But here’s the thing nobody tells you: