"This is the death of my professional identity."

That was David Grandinetti's first reaction after his team decided everyone—from designers and product managers to executives—should be contributing directly to the codebase with AI. Grandinetti, a Distinguished Software Apps Engineer at Yahoo, had spent decades writing software. Watching AI generate production code felt less like adopting a new tool than confronting an existential shift.

But over time, Grandinetti realized that solving difficult problems had always been what he loved most about engineering. "I like solving problems with elegant code, but more than that, I just like solving problems,” he says. “Now, I can solve problems that I never would have thought to attack before."

That realization sits at the heart of one of the biggest questions in software engineering today. AI has dramatically reduced the cost of producing software, forcing the profession to reconsider what, exactly, makes an engineer valuable. Inside Yahoo, the role is moving to a higher level of abstraction. Engineering value increasingly comes from defining problems, evaluating solutions, and improving the way teams build software.

"Generating code is now the cheapest part of the process,” says Yahoo COO Matt Sanchez. “It was the long pole, and now it's easy to write code, throw it away, and try something different. The real work is moving to specification and evaluation—to thinking through what you're trying to accomplish rather than the implementation layer. That will play to the strengths of the very best engineers."

The real craft of engineering‍

Much of the discussion around AI has focused on code generation. Yahoo CTO Lee Zen believes AI is bringing the rest of engineering into sharper focus.

"Even before AI, the amount of time someone spent coding was probably something like 20%," Zen says. "The job is a lot more about design and testing and operations and everything else that goes around how to actually build and deploy an application."

The job is a lot more about design and testing and operations and everything else that goes around how to actually build and deploy an application.

That broader perspective changes how he views AI's impact. Writing code is only one step in building software. AI will most certainly alter the non-coding portions of an engineer’s job the way it has coding. But engineers still define systems, balance competing priorities, and make decisions that shape how products perform, scale, and evolve.

Every engineering decision creates a new set of constraints. A solution that's faster may be harder to maintain. One that's easier to extend may cost more to operate. Improving one part of a system often means compromising another.

For Zen, that's the real craft of engineering. Performance, reliability, maintainability, cost, security, and future flexibility all compete for attention, and different engineers may arrive at different solutions. The strongest engineers consistently make decisions that balance competing priorities in ways that best serve the product, the team, and the business.

Building better builders‍

The challenge for engineering leaders is making technical judgment an organizational capability. That's the transition Kat Styons, Senior Director of Software Development Engineering at Yahoo, sees underway. As AI accelerates software development, helping every engineer make better decisions creates more value than helping a single engineer work faster.

That requires capturing successful workflows, building automated quality checks, and creating systems that allow engineering judgment to spread from one project to the next. Over time, those investments compound. Every new workflow, framework, and development harness becomes part of the organization's engineering system, giving the next team a stronger starting point than the last.

"You almost need to demo not just what you built, but also how you built it," Styons says. "For a little while, the product that you're building is also going to be your way of working."

For a little while, the product that you're building is also going to be your way of working.

In Styons' view, the implications extend beyond engineering workflows. As the cost of producing software falls, the new limiting factors become the judgment of the people on a team and how quickly an organization can learn from each success and failure. 

That philosophy also shapes how Styons thinks about hiring. Curiosity and intellectual courage have moved to the top of her criteria because engineers increasingly need to question AI's output, develop informed opinions, and guide coding agents rather than simply accept their suggestions. Engineering leaders also have to create new ways for junior engineers to build judgment and for experienced engineers to share it.

"We have a responsibility as leaders to figure out what the new career ladder is going to look like," Styons says, "because the rungs that we climbed are no longer there."

Key takeaways

Engineering value is shifting toward judgment. As AI reduces the cost of implementation, engineers create value by defining problems, evaluating solutions, and making the tradeoffs that shape products.

Great engineering organizations make judgment reusable. Shared workflows, automated quality checks, and development systems allow hard-won expertise to compound across teams instead of remaining with individual engineers.

The new bottleneck is organizational learning. As code becomes easier to generate, competitive advantage increasingly depends on how quickly teams learn, adapt, and turn experience into better ways of building.

‍