Tom Archer had already spent years coding as a self-taught developer before he joined Microsoft, where for the last two decades he has worked as a senior AI engineer. Then he made a decision that many did not see coming. He went back to school to study computer science. As the AI wave came to town, he decided to go back again to get his master's.
With the rise of vibe coding tools like Loveable and Replit, many young tech misfits have landed coding jobs. In February, Spotify co-CEO Gustav Söderström bragged on an earnings call that the best developers at the music and podcast streaming company “have not written a single line of code since December.”
But Archer, who has been working as a senior software engineer since 1985, says things are changing. "In today's competitive AI landscape, that theoretical foundation matters. It distinguishes engineers who can use tools from those who can build them,” he told Techloy. "If you have an AI degree, you stand a higher chance of getting hired. That is just the reality," Temitope Adejolu, an AI Software Engineer who works at Interswitch, said.
The debate presents the latest generational divide between the early builders of what we now call the World Wide Web and the younger generation of builders, irritated by gatekeepers and busting with energy to break things.
Kosi Ashara, a UK-based AI engineer with a master's in artificial intelligence, says even a degree is not needed. "But people would say I'm saying that because I have one,” she said. For her, a portfolio of your work was all you needed to advance your career.
But many of the young engineers I spoke to who argued that portfolios matter more than a degree hold one in computer science, AI, or a related field. "Build like no tomorrow, because in this age of AI, it's your work that will speak for you," Grace Ajagbe, lead AI consultant at Automaticah, told me.
Archer is the only engineer in this group who built his career first and went back for formal education. "Don't let the lack of a degree stop you from starting, but don't dismiss its value either," Archer said. "What gets you through the door at most companies is demonstrated ability. But a degree matters more than some people want to admit, especially at research-oriented organisations and for long-term career growth.”
For him, he sees very little pathway for an AI engineer who does not learn the basics of the trade. "Python is the lingua franca of AI engineering,” he said. “The ecosystem around it—PyTorch, Hugging Face, LangChain—is where most of the real work happens."
Damilare Odedina
Even the younger engineers agree with him. "Brush up on Python or JavaScript, API and JSON. These are the foundations," Ajagbe said.
But the idea that vibe coding tools can lead to a sustained career as a developer or AI engineer is not without merit. However, accelerated growth in the industry has drastically altered its focus as investors begin to demand mainstream products from startups.
"In 2023, it was about prompt engineering, about how to give models the right prompt. Then it evolved into RAG, retrieval augmented generation, storing your information in a vector database. Now we are in the world of MCP agents and agentic AI. It's not going to stop, because people are always going to come up with different ideas to use these LLMs better," Ashara said.
Hiring managers at tech companies have now been tasked with finding engineers who understand all three stages. But Ashara said even with this mandate, there might be a way out for people willing to do a lot of research. "Try to understand the concepts of how all of these things work, and try to use these LLMs and agents to build different stuff," she said.
Archer sees the refusal of vibe coders who get a foot in the door and fail to go back and learn the basics of the trade as the “worst habit” a beginner can have. He referred to it as “cargo cult programming.”
"There's a temptation to skip the fundamentals and jump straight to calling APIs and importing pre-built models," he explained. "That works until it doesn't. Without understanding how things work under the hood, when something breaks or behaves unexpectedly, you're stuck."
Ajagbe also echoes what he said, albeit with the caveat that vibe coding is fundamentally good. “In the long run, don't just vibe with AI, learn judgment, learn problem solving. That is what will make you different," she said.
Vibe coding has made solving problems easier, giving almost every applicant a high-quality portfolio and example to show. In the past, when just showing the problem that was solved was enough, hiring managers are now looking for more work-throughs.
Archer says that what will impress hiring managers is “when someone identifies a genuine problem, chooses an appropriate approach, builds a working solution, and can articulate the tradeoffs they made along the way."
Archer, who helped run the blog CodeGuru, said that starting one helps hiring managers understand the candidate is more than just a vibe coder.
"Build a habit of learning in public. Start a blog, contribute to open source, and engage with the community. Your ability to learn, adapt, and communicate what you know is more valuable than any specific skill you have today," he said.
"Reinvent yourself always. That's how to stay relevant,” Ajagbe said.
Ogbonda Chivumnovu
