This video features an in-depth conversation with Gergely Orosz, author and founder of The Pragmatic Engineer, covering the transformation of software engineering in the age of AI and how to navigate your career through it. He draws on his own experience growing through chaos at Uber, shares what separates successful engineers from the rest, and digs honestly into what it feels like when AI starts replacing the coding skills you spent years building — and what capabilities will matter going forward. The conversation offers practical advice and genuine insight for anyone thinking about how to grow as an engineer in the AI era.

1. Starting a Tech Career and Life at Uber

Gergely Orosz started coding in PHP back in high school, but when he enrolled in a computer science program at university, theory-heavy coursework nearly killed his enthusiasm. Struggling through subjects like C and 3D graphics was frustrating, yet it was through those hard subjects that he ultimately found his passion for software development. He reflects that this foundational knowledge later proved invaluable for understanding the relationship between LLMs and GPUs.

"I understand that LLMs are a lot of matrix transformations, and 3D graphics is all about matrix transformations and GPUs, and matrix transformations can be parallelized and GPUs are good at parallelization — so that's why this works."

After consulting in Hungary and the UK, and building experience at companies like Skype and Skyscanner, he finally joined Uber in 2016 — at the time one of the hottest companies in the world. From the outside, Uber looked incredible. From the inside, it was chaos held together with duct tape and Google Sheets.

"From the outside, Uber looked great. But inside, everything was connected with duct tape and Google Sheets."

He explains that Uber's adoption of a microservices architecture was actually a deliberate strategy for managing explosive headcount growth — a small number of senior engineers built infrastructure on platform teams, while junior engineers shipped independent services and learned fast. He notes that many companies copied the microservices approach while completely missing the strategic rationale behind it.

Despite the chaos, Gergely says he learned an enormous amount. He recounts a telling story: an engineer who transferred in from Google couldn't adapt to Uber's disorder and quit within a week. His manager at the time drove home the importance of transparent communication — be honest about problems, don't hide them.

"I think it was good this person left. It was better for them, but they shouldn't have joined in the first place."

Uber insisted on building its own data centers and maintaining a proprietary tech stack, which he sees as a genuine advantage in hindsight: all that in-house building gave engineers opportunities to develop deep knowledge and experience across the stack.

2. Moving into Management and What Makes Engineers Succeed

Gergely's transition into management was largely accidental. He had applied to Uber as an iOS developer, was offered a backend role at the interview, and then — right before joining — was asked to work on Android instead. Uber was in the middle of a massive project to completely rebuild its app in three months. When Gergely saw the project was heading for disaster, he said so directly and proposed a path forward, which naturally put him in the role of project lead.

"I said, 'I think our project is a disaster.'"

After steering that project to success, he was formally made a manager. At the time, moving into management at Uber was essentially a lateral move — no meaningful pay bump — but Gergely chose it for the learning opportunity the new role represented.

The trait he most consistently observed in successful engineers was being product-minded: going beyond writing good code to deeply understanding business goals and user needs, and proactively contributing to product improvement.

"They understood what we were trying to do. They took the initiative to test what competing apps were doing. I would call them product-minded software engineers."

These engineers actively communicated with teammates, met regularly with product managers to stay on top of the business, put forward feature ideas and improvements, and had the courage to push back on unnecessary work with a clear "no."

His advice for engineers thinking about moving into management: genuinely enjoy mentoring, and find real opportunities to try the role — ideally in a safe environment like Uber's apprentice manager program. He also notes that engineers who try management and return to individual contributor roles often become better engineers because they understand and collaborate with managers more effectively.

3. How AI Is Changing the Engineer's Role — and What to Do About It

Gergely is candid about the fact that AI writing code triggered real doubt about the value of the coding skills he had spent over a decade building.

"With AI writing code, it almost feels like everything I worked on for more than ten years to become a really good programmer has nearly vanished."

He describes how coding interviews used to measure an engineer's grit and raw intelligence — signals that mattered precisely because coding was hard. Now that AI puts coding within reach of almost anyone, that signal has eroded. He admits to genuine surprise at discovering that the latest LLMs produce code as good as — or better than — what he writes himself.

On the question of expertise in an AI-driven world, he argues that the 10,000-hour rule may no longer apply in the same way — becoming proficient with AI tools takes far less time. But that doesn't mean learning and effort stop mattering. The professionals who will succeed are those who use AI as a supporting tool while staying in the driver's seat — people who take ownership of understanding and solving problems rather than just accepting AI output.

"You could argue that AI knows everything so you can skip learning, but I think the people who will succeed in every profession are those who put in the effort, understand things, and use AI as a tool."

He stresses the importance of critical thinking about AI-generated information, and cites research showing that beginners tend to trust AI answers uncritically — a pattern he finds alarming. He shares a personal example: legal advice he got from AI turned out to differ from what an actual lawyer told him. The lesson was the value of real-world context and professional authority.

"As a beginner, I would have trusted ChatGPT completely. But it turned out information was missing, there was no awareness of the real world, and there was no training data on these kinds of issues."

In that light, he argues that no matter how capable AI becomes, building business-critical software still requires a human who can be held accountable when things go wrong.

4. Career Trends and the Skills That Will Define Future Engineers

Several notable career trends are reshaping software engineering right now:

• Smaller teams: The classic "two-pizza team" of 5–10 people is shrinking toward a "one-pizza team" of 2–4.
• Fewer engineering manager roles: Manager positions are being cut; the emphasis is shifting toward more hands-on managers. Many companies are moving back toward a flat structure with a small number of directors and a larger number of tech leads.

"The engineering manager career is becoming less common and less lucrative."

He challenges the popular advice that job-hopping is the fastest path to higher pay and faster promotions — pointing out that this was only ever true in a hot market. If your goal is a leadership position long-term, staying at one company for three or more years and demonstrating the impact of your decisions matters far more.

"Frequent job-hopping becomes a red flag when you want a leadership position or to be considered a very senior individual contributor."

He also emphasizes that doing great work alone isn't enough — you need to make your contributions visible. Especially in large organizations, advocating for yourself is essential to being recognized and fairly evaluated.

Quoting Mitchell Hashimoto, he makes a case for the value of engineers with empty GitHub profiles. No flashy side projects or open-source contributions — just engineers who show up and give 100% to the work in front of them, achieving excellent outcomes efficiently. In his experience, these engineers are often the best ones.

"The best engineers often have empty GitHubs. They actually leave at 6pm on the dot to pick up their kids. They come in at 8am and work really focused for eight hours."

The three most important capabilities for software engineers in the AI era:

1. Adaptability: The ability to respond flexibly to a constantly changing technical landscape.
2. Curiosity: A genuine drive to explore new technologies and keep learning.
3. Leaving your ego at the door: An open, non-defensive posture toward new approaches — willingness to let go of familiar ways of doing things.

He draws on Grady Booch's concept of the "leap of abstraction" to frame the current moment: just as the shift from assembly language to high-level languages once seemed threatening to programmers who had mastered the old way, the shift to AI-assisted coding is a similar inflection point. Programmers who rejected early compilers got left behind; those who embraced the new tools thrived. And the people who understood both the old and the new became the most powerful of all.

"Coding might be something a machine can do, but building trustworthy software — and building it fast — is not going away."

He holds up Boris Churnney, creator of Claude Code, as a model engineer for this new era. After a motorcycle accident left Churnney able to code with only two fingers, he taught himself functional programming rather than working around his limitation in conventional ways — an example of willingness to discard old habits and learn something genuinely new.

5. Rapid-Fire Questions and Closing Thoughts

To close, Gergely runs through a set of rapid-fire questions and shares his views:

• Startup vs. Big Tech: "Startup. And then a bit of Big Tech experience on top — that's the ideal combination." He recommends doing both.
• Generalist vs. Specialist: "Generalist these days. With AI, you can become a specialist in whatever area you want." His view: AI enables a new kind of generalist who can go deep when needed.
• Remote vs. In-person: "In-person. The more time you spend face-to-face, the stronger the human relationships you build — and those relationships are what make remote work possible later."
• Backend vs. Frontend: "These days it's all full-stack. Backend is still a bit more demanding. And frontend still requires a good eye, even with AI — AI is much better at backend."
• Python vs. Java: "Neither." If forced to choose he'd pick Python, but his real preference is TypeScript — he loves C's depth, but TypeScript gives him a strong type system with genuine flexibility.
• Tabs vs. Spaces: "Tabs — no contest."
• Async communication vs. meetings: "I want to say async, but if you need a meeting, in-person is better."
• Whiteboard interviews: "They'll come back. Once AI is everywhere, remote interviews become useless. Whiteboard interviews are important for figuring out who can actually think on their own."

On the final question — "Will AI replace software engineers?" — his answer is an unequivocal "No."

"No, AI will complement software engineers. Software engineers won't lose their jobs to AI — but they will lose them to software engineers who are very comfortable with these tools, know when to use them, stay curious, and keep up with what's happening."

AI is a tool. Engineers who wield it skillfully will displace those who don't. The imperative for every engineer is therefore continuous learning and adaptation — meeting the new standard that the AI era demands.

Conclusion

Gergely Orosz's core argument is that software engineering in the AI era demands far more than the ability to write code. It requires business understanding, problem-solving judgment, and a relentless commitment to learning and adapting. AI will accelerate the coding layer of the work, but the human engineer's role — building trustworthy software, thinking critically, and taking responsibility — will only become more important. The engineers who thrive will be the ones who use AI to amplify their own capabilities, approaching change with flexibility, curiosity, and an open mind.