Hamza Mostafa's path into frontier artificial intelligence did not begin with a polished resume, a famous research lab, or a direct route into Silicon Valley.
Before he became part of OpenAI's first intern cohort, Mostafa was a University of Waterloo computer science student trying to break into software engineering through the same uncertain process many students face: applying widely, getting rejected, and looking for any opportunity that could become a starting point.
His first software internship was unpaid. It was not at a major technology company or a well-known AI lab. It was with a very early software project led by a Waterloo senior.
"It was not glamorous, but it gave me a starting point," Mostafa said. "I could not get a paid software role at first, so I took the opportunity I had and tried to turn it into the next one."
That mindset eventually led him through roles at Whatnot and Verkada before he reached OpenAI. But Mostafa says the path looked much cleaner from the outside than it felt while he was living it.
"I was rejected many times," he said. "I cold-emailed recruiters. I asked friends to practice interviews with me. I applied early. A lot of it was just taking the next imperfect opportunity seriously."
Born in Cairo, Mostafa moved to Canada when he was 10. For years, he expected to pursue medicine or forensic science. Computer science became interesting to him when he realized software was becoming one of the highest-leverage ways to build products, solve problems, and participate in the future of technology.
Once he made that switch, he became intentional about putting himself in more demanding engineering environments. After his first unpaid role, he kept trying to move closer to teams with faster execution, higher technical bars, and more ambitious problems.
That progression eventually brought him to OpenAI at a rare moment: the company was preparing its first intern cohort.
Mostafa applied the day the applications opened after a friend sent the link in a group chat. Because the cohort was new, there was little public information about what OpenAI wanted from interns or what the interviews would look like.
He expected the process to center heavily on artificial intelligence. Instead, he said the interviews focused on core software engineering: algorithms, system design, speed, clarity, and behavioral judgment.
"The surprising thing was that it was not all deep AI questions," he said. "The bar was more about whether you could think clearly, communicate well and move quickly under pressure."
That pace continued once he joined.
Mostafa said OpenAI felt faster than he expected from a large technology company. He shipped code on his first day and found that the environment required interns to learn quickly, ask strong questions, and take ownership without waiting for every detail to be explained.
"At OpenAI, I learned that speed matters, but judgment matters even more," he said. "The best engineers were not just moving fast. They were honest about what was working, what was failing and what needed to be tested more carefully."
That lesson has shaped how Mostafa thinks about the future of AI engineering.
Much of the public conversation around artificial intelligence focuses on model capability. More capable models can write code, summarize information, retrieve data, use tools, and help complete increasingly complex tasks. But Mostafa believes the next phase of AI will be defined not only by intelligence, but by whether people can trust these systems in real use.
AI agents are especially important in that shift. Unlike simple chatbots, agents can take actions. They can navigate tools, execute tasks, and make decisions inside workflows. That creates new opportunities, but it also introduces new risks.
"A chatbot giving a bad answer is one thing," Mostafa said. "An agent taking the wrong action is a much bigger problem."
For that reason, he believes young engineers entering AI need to think beyond demos. It is easy to build something that looks impressive once. It is harder to build systems that work repeatedly across messy, real-world situations.
Reliable AI systems need evaluation, testing, transparency, oversight, and a clear understanding of when a human should step in. They also need engineers who can recognize when an AI-generated answer looks plausible but is actually wrong.
That is why Mostafa describes the next generation of engineers as needing to become AI-native, but not AI-dependent.
To him, an AI-native engineer is not someone who blindly accepts model outputs. It is someone who uses AI tools to move faster while still understanding the underlying system. AI can help write code, generate tests, explore ideas, and reduce tedious work. But it can also create subtle errors or give users confidence in outputs that have not been properly verified.
"AI can make a good engineer much faster, but it can also make an inexperienced engineer overconfident," he said. "The fundamentals still matter. If you understand the code, AI helps you move faster. If you do not understand the code, AI can create the illusion of progress."
Mostafa uses AI coding tools regularly now, even though he was more skeptical of them earlier in his career. He said the shift happened when the tools became useful enough for serious engineering work, not just simple experiments.
Still, he believes students should continue studying computer science deeply. Algorithms, systems, databases, networking, and software design remain important because they give engineers the foundation to judge whether AI-generated work is correct.
His advice to students trying to enter AI is practical: build real projects, use AI tools every day, practice interviews with friends, apply early, cold email when appropriate, and get close to serious people solving serious problems.
Moving to San Francisco also changed his perspective. Being around ambitious builders helped him understand the pace of the industry and the kinds of problems leading teams were working on. One of his later opportunities came through someone he met in person.
But he is careful not to make the story sound too simple.
Before entering frontier AI, Mostafa worked jobs far from the world of advanced technology, including as a janitor. He later shared that part of his story publicly because he believes people often see only the final outcome, not the uncertainty that came before it.
"When I shared that part of the story, it resonated because a lot of people understood the uncertainty," he said. "People see OpenAI on a resume and assume everything was straightforward. It was not."
For Mostafa, OpenAI showed him what the frontier of AI looks like. But the path there taught him something equally important: progress often comes from taking the next opportunity seriously before feeling fully ready.
"The future of AI will not belong only to people who can build impressive demos," he said. "It will belong to people who can turn powerful technology into systems that work reliably in the real world."
