You can’t search LinkedIn for judgment

Companies keep asking me to find engineers with judgment.

Sometimes it’s ownership, curiosity or commercial sense. They want someone who can deal with ambiguity, challenge an assumption and work out what actually needs building.

You absolutely should be hiring people with all of the above. It just makes my job more interesting.

Most engineers are using AI to work across areas that would previously have taken them much longer to understand. They can explore an unfamiliar codebase, test an idea and produce working software without knowing every part of the stack beforehand.

That makes judgment more important. Producing something is getting easier. Deciding what to produce, whether it is any good and when the obvious answer is wrong still requires someone who understands the problem.

LinkedIn is good at finding visible experience. It can tell me where somebody has worked, roughly how senior they are and whether they have used a particular language or framework.

It is considerably less useful when the brief is to find someone who will use AI well without trusting everything it gives them.

LinkedIn isn’t the only place to look. GitHub, meetups, technical communities and referrals can all offer clues that a profile won’t.

But they don’t give you a clean answer either. A public project shows what somebody built. It doesn’t show why, or which parts they understood. A referral tells you that someone rates them, but not whether they are right for this particular problem.

AI makes this harder again. More of the profiles I see now mention agents, copilots and AI-assisted development. That tells me somebody uses the tools. It doesn’t tell me how they use them.

What to ask instead

The useful evidence only comes out when I speak to them.

On a first call, I want to understand what they were trying to achieve, what they let the tool do and what they checked themselves. I want to know when they rejected its answer, whether it helped them understand the problem and what happened when the work met real users.

The same applies beyond AI. What was unclear when they started? What did they personally decide? Which approach did they reject? What changed because of their work?

Those questions tell me more than asking someone whether they are comfortable with ambiguity. Almost everyone interviewing for the role will say yes to that.

This connects with my last Field Note about engineers becoming broader while hiring processes continue to reward specialisation. If AI helps people move into unfamiliar areas more quickly, it becomes harder to infer their ability from a list of technologies they have already used.

That doesn’t make specialist knowledge irrelevant. There are roles where mistakes are expensive and the company needs someone who already understands the problem deeply. Using AI confidently is not the same as using it well, and it certainly doesn’t replace experience in every role.

The problem is when a company says it wants judgment, ownership and learning speed but runs a search designed mainly to find familiar titles and technical keywords.

I don’t think the answer is replacing those keywords with “curious” or “high agency.” Once companies start asking for those qualities, everyone learns to add them to a profile.

You still have to look at the work and ask better questions about it.

That is harder to turn into a filter, but it gets much closer to what companies actually need.