Estimated reading time: 5 minutes

Key takeaways:

• Engineers are staying indispensable by doubling down on skills like product sense, architectural design, and deep customer empathy that AI cannot yet replicate.
• Rather than resisting, top developers use LLMs and agents to automate rote tasks, allowing them to focus on high-level problem-solving and system reliability.
• Technical rigor remains vital. Human judgment is necessary to verify AI-generated code and ensure systems don’t become fragile.

As layoffs rise and AI reshapes the profession, some developers are doubling down on the craft and learning how to be indispensable.

For software engineers, career uncertainty has become the norm. This year’s layoffs have already surpassed the start of 2025 by more than 20%, according to data from layoffs.fyi. 

CEOs at Spotify, Atlassian, Block, and a host of other companies have been happy to attribute labor efficiencies to AI – regardless of whether the tools are truly capable of replicating human engineers’ output.

Facing a labor market that looks more and more like a high-stakes game of musical chairs, some engineers have opted to flee big tech or switch careers altogether. 

Many others, however, are standing their ground and keep doing the jobs they enjoy and are good at. And they’ve made plans to remain indispensable as the industry continues to evolve.

Staying in the game

Clive Dsouza, a senior software engineer for Intuit CreditKarma, can map his entire career against a timeline of technological advancement. He started out two decades ago as a backend engineer working in Java, J2EE, and EJB, before switching to early frontend web technologies such as JQuery, JavaScript, HTML, JSP, Servlets, and JSF. From there, Dsouza moved into modern frontend frameworks: BackBone.js, Angular and, most recently, React. 

As Dsouza sees it, staying on top of new programming languages, frameworks, and platforms not only comes with the territory of building software, but is part of the job’s appeal. “I like the problem-solving aspect of my work, as well as continuously learning in order to keep up with the ever-changing tech industry,” he says. 

Roman Martynenko, founding engineer at Henry AI, similarly plans to stay put. “I’ve been naturally drawn to computers and new technology since I was a kid,” he says. Following his passion wasn’t always easy; growing up in central Ukraine during the late ’90s and early 2000s, computers were expensive and online connectivity was both difficult and costly to attain. “Most people didn’t really understand what the internet was or where it was going, which somehow made it even more compelling,” Martynenko recalls.

When Martynenko built his first websites with a friend, there were no online courses or tutorials – let alone LLMs – accessible for reference. “We were learning from books, experimentation, and intuition. Just trying things, breaking things, and getting that ‘Wow, I made this work’ feeling.”

Now, the make-it-work ethos is one Martynenko must apply within a new set of parameters. “The bar is shifting from being able to implement a well-defined ticket to figuring out what to build, why, and measuring how well it succeeded. Product sense, design awareness, and the ability to deeply understand customer needs are becoming core engineering skills.” 

That means leaving the code editor to observe real user behavior, talking to customers, and partnering across sales, ops, and design functions to make decisions that create measurable business value, Martynenko explains.

Gabriela Moreira’s affinity for software also began in childhood. An avid gamer with an unreliable computer, Moreira realized that she enjoyed finding ways to make her games playable almost as much as she liked playing them. Later, as a university student pursuing computer science in her native Brazil, she found herself drawn to computer theory and compilers. “I loved finding solutions to hard problems, and I realized those same challenges show up in distributed systems,” recalls Moreira, now the lead developer at Quint, a specification language based on TLA+ that adds type checking and developer tools.

Problem solving continues to be Moreira’s favorite part of her job. “I love to ask ‘What if?’” she says. It’s an approach she applies across functions – whether designing language and user interfaces, drafting social media posts, or chatting with customers. But software is still her favorite use case for devising new solutions: “It’s where I can get the most precise answers to my what-if questions, and I get to either celebrate a win or sit down and think about another approach.”

Sharpening the toolkit

Engineering holdouts are not naive optimists. Rather, they understand that the fundamental traits of a good engineer – adaptability, judgment, curiosity, and problem-solving acumen – can serve as protective armor against the whims of a shaky job market. In the near term, that means learning to build with AI. 

Dsouza now relies on an array of generative AI programming tools to support his daily workflows, using Claude Code and Cursor with Claude Opus to generate code and running agents on his Mac Mini with OpenClaw to automate personal projects. “I am automating my tasks so that I can focus on more important decisions, especially around design and architecture, and learning as much as I can so that I can stick around in this changing industry,” he says.

Martynenko is also working to build his technical acuity with LLMs as engineering tools, while making sure to stay disciplined in ensuring quality. “LLMs are a multiplier, not a replacement,” he says. “Without fundamentals, you can ship faster in the wrong direction or produce fragile systems that break under real-world constraints. And being able to use LLMs effectively is now a must-have because it dramatically accelerates iteration, debugging, and exploration when used with strong judgment.” His primary objective, for both his own work and that of his current and future team members, is the ability to effectively use AI without outsourcing thinking. 

Moreira is also reaping the benefits of AI acceleration. “I can honestly say I’ve never been more excited about the future of this industry,” she says. “While many view the shift to AI and vibe coding as a potential hurdle, I believe it simply makes my work that much more impactful. Now more than ever, as AI generates massive amounts of code, we need ways to actually verify the solution is correct and properly implemented. It actually feels like the industry is now catching up to the problem I’ve spent the past several years working on.” 

Engineering holdouts share a special thrill from finding solutions to problems that deliver real, useful results. For Martynenko, the process of experimentation, discovery, and eventual awe at finding the right approach is still as riveting as it was more than 20 years ago, when he was just a boy in Ukraine tinkering on the web.

“What I still love about the job is that same loop: taking something vague, turning it into something real, and watching it create value for someone,” Martynenko says. “It’s a mix of creativity and engineering rigor, and you get immediate feedback from the real world.”

New York • September 15-16, 2026

Speakers Gergely Orosz, Will Larson and Frances Thai confirmed 🙌

Explore LDX3