Hiring Engineers in 2025: What Actually Works

Entry-level engineering jobs have dropped 60% since 2022. That is not a slow decline. That is a collapse. At the same time, senior AI talent demand is running at 3.2 to 1 against supply. Every company wants the same 200 people. Nobody is building the next generation.

I have been hiring engineers for over 20 years. I scaled a team from 12 to 200+ at a global logistics company operating across 130 countries. I built a $5M consulting practice where every single hire mattered to survival. I now lead a 35-person team at a retail data company. I have got it wrong many times. Here is what I have learned.

What I look for

I do not care where you went to university. I care whether you can think through a problem I have not seen the answer to either.

The best engineer I ever hired had no computer science degree. She had studied biology. She taught herself Python to automate lab data. In her interview she asked more questions than I did. That curiosity is what separates good engineers from great ones.

Credentials tell me what someone studied five years ago. Problem solving tells me what they will do tomorrow when something breaks at 2am and the runbook does not cover it.

I look for three things. Can you break a vague problem into smaller clear problems? Can you explain your thinking out loud without jargon? Do you ask why before you start building?

That last one matters more than people think. The engineers who cost me the most were the ones who built the wrong thing fast.

How I run interviews

No whiteboard puzzles. I have never once needed someone to reverse a binary tree on the job. Those tests measure how much interview prep someone did, not how they work.

Instead, we use real problems from our codebase. Not trick problems. Actual bugs we fixed last month or small features we shipped recently. We give candidates the problem, the relevant code, and a laptop. They have 45 minutes. They can use Google. They can use Stack Overflow. They can use AI tools. Just like real work.

What I am watching is not the answer. I am watching how they read unfamiliar code. Do they start by understanding the system or do they jump straight to typing? Do they test their assumptions? Do they ask clarifying questions?

We also do a 30 minute conversation with someone from the team who would work with them daily. Not a panel. One person. Relaxed. The question is simple: would I want to debug a production issue with this person at 11pm on a Friday?

How I keep people

I have lost good engineers to companies offering 20% more money. I have also kept engineers who turned down 40% more elsewhere. The difference was never the office or the perks.

People stay when they own something real. Not "ownership" in an HR slide deck. Actual ownership. This service is yours. You decide how to build it. You are accountable when it breaks. You present the results to leadership.

The engineers I have kept longest all say the same thing: they stayed because nobody told them how to do their work, only what outcome was needed.

The other thing that keeps people is growth that matches where they want to go. Some engineers want to become architects. Some want to lead teams. Some want to stay deep in code forever. All of those are valid. The mistake is assuming everyone wants the management track.

Ping pong tables and free lunch do not retain anyone. Interesting work and genuine autonomy do.

The junior pipeline problem

This is the part that worries me most.

If entry-level postings are down 60%, where are the senior engineers of 2030 going to come from? They do not appear fully formed. It takes 4 to 5 years of working on real systems, making real mistakes, and learning from them.

Every company wants to hire seniors. Nobody wants to train juniors. The maths does not work.

At my last three roles I have always kept a junior intake even when budgets were tight. It is slower in the first year. By year two those juniors are contributing more than some mid-levels because they learned on your systems with your patterns.

The industry needs to stop treating junior hiring as charity. It is an investment with a clear return. You just have to be patient enough to wait for it.

AI tools have made this worse, not better. Companies think they can replace junior work with AI code generation. But someone still needs to understand the system well enough to review what the AI produces. That understanding only comes from doing the work yourself first.

The one hire I regret most

Three years ago I hired a brilliant architect. Top of every technical assessment. Deep expertise in distributed systems. Exactly the skills we needed.

Within six months the team was miserable. He was technically right about most things and made sure everyone knew it. He reviewed code like he was marking exam papers. Junior engineers stopped asking questions because they were afraid of looking stupid.

I had hired for skills and ignored every signal about how he treated people. The warning signs were there in the interview. He interrupted. He did not listen to follow up questions. He gave answers designed to impress rather than communicate.

I should have trusted my gut. Instead I was too impressed by the technical depth. It took me too long to act and by the time he left, two good engineers had already gone.

Skills can be taught. Decency cannot. I will never make that trade again.

What I would tell hiring managers right now

Stop copying FAANG interview processes. You are not Google. You do not have their candidate volume or their problems.

Stop writing job descriptions that list 15 required technologies. You need someone who can learn, not someone who already knows your exact stack.

Hire at least one junior for every five seniors. Your future depends on it.

Speed up your process. The best candidates are gone in 10 days. If your hiring loop takes 6 weeks, you are only interviewing people nobody else wanted.

And when you find someone good, trust them. Give them real work. Get out of their way. That is what actually works.