The Engineering Squeeze

Imagine a spectrum: on the left, the most “business” concepts possible - perhaps an MBA. On the right, the most technical of technical, perhaps the very transistors that form the logic gates that our code dances through. Software engineers have traditionally occupied a slice somewhere in the middle - they take needs from a business stakeholder, write some code that gets compiled and shipped off to the “server” somewhere.

But with LLMs able to generate code faster than any human can, where does that put the humble software engineer? From my vantage point, it seems like we are being forced outward, further from the idea-into-code role in both directions. As we move to the right, we need to be able to understand and troubleshoot systems more holistically. Design, architecture, properties of complex systems become increasingly important. On the left, we more and more must reconcile with business problems.

It’s this side of the spectrum that I’ve been finding more interesting recently. After all, software is a means to an end. Instead of asking “what is the exact shape of the data this API exposes?” or “What color do you want the button?”, we must now more than ever be asking, “what is the problem we’re trying to solve with this product?” “Will this bring meaningful value to the business?” “Is this the right thing for us to focus on right now?” “How will we know?” “What can we do to find out?”

We all hear a lot about how AI is going to change things, good and bad. I feel that much of the negative sentiment is fueled by fear of being replaced, which may ultimately be the fate of some folks who fail to adapt. But I’d like to offer an alternate way of looking at our current situation - most software engineers got into the field because they enjoy solving problems, and from my seat, we are not short on those. While it’s clear from our human condition that not every problem can be solved with software, I don’t see us running out of problems that we can solve with technology anytime soon. AI is an immensely powerful tool, but it’s still just that - a tool: one that we can choose to wield to help those around us.

If you have been a code monkey for the past three decades and you’re feeling nervous about your career, or if you just graduated college with a degree in a field you’re not sure wants you, don’t focus on grinding leetcode. Push left on this spectrum - focus on learning to ask questions that lead to solving problems. I hired several engineers recently at my company. When I was interviewing, I was looking for people who liked solving problems. Keep your chin up.