How You Actually Grow as an MLE

Most people think levelling up is about getting better at coding.

It’s not. Or at least, that’s not the interesting part.

Here’s how I actually think about the levels and what changes at each one.

L3: You execute

Someone tells you what to do and you do it well. Try this experiment, model it after this existing one, here’s the design: go implement it.

Blocked? You flag it immediately and ask your tech lead.

Your job is precise execution and clear communication about where you’re stuck. You’re not expected to figure out what to build. You’re expected to build it correctly once someone tells you.

That’s not a criticism. It’s the right way to start. You’re learning the system, the codebase, the culture, how decisions get made. Absorb everything.

L4: You operate independently

No more hand-holding. You are the team’s coding horsepower.

You pick up tasks without being told how to approach them. You start writing design docs, maybe with some support from a senior engineer the first few times. You’re not just flagging problems anymore. You’re showing up with proposed solutions.

This is where most ML engineers spend a significant chunk of their career. And honestly? Being a great L4 is genuinely valuable.

Teams need people who can just execute at high velocity without creating overhead for everyone around them.

What comes after L4 is where it gets interesting and where most people get it wrong.

The levels, the paths, how I personally think about my own trajectory, and the archetype that’s worked for me.