Last week I looked into applying for a job at Terafab, the semiconductor megafactory that Tesla, SpaceX, xAI, and Intel announced in Austin. I read through the roles they were hiring for. Process engineers. Chip architects. Hardware specialists. People who work with silicon, not JavaScript. My stack is Next.js, React, TypeScript, MongoDB, Swift. I am based in Istanbul, not Texas. I am eighteen years old. I do not have a work visa. I do not have a degree.
The sensible conclusion arrived quickly: not this one. Not yet, possibly not ever for that particular operation. But the exercise clarified something I had been putting off thinking about. There are doors that cannot be opened by shipping products alone, no matter how many users those products serve. Some doors require credentials.
So I am applying to university. Computer science, with a focus on machine learning, for 2027 entry.
This feels stranger than it probably should. I have been writing production software since I was twelve. I run a company with four divisions, staff across countries, clients who rely on infrastructure I helped build. I have shipped products used by millions of people. And now I am studying for the SAT, a standardised test designed to measure whether I am ready for higher education.
The absurdity is not lost on me.
But absurdity and necessity are not mutually exclusive. I have read Andrew Ng's CS229 lecture notes. I understand backpropagation and gradient descent. I can follow a conversation about transformer architectures without getting lost. What I do not have is the mathematical depth that separates someone who uses machine learning from someone who advances it. Linear algebra at the level it is actually practised. Optimisation theory. The kind of probability and statistics that makes new research possible, not just comprehensible.
I can teach myself many things. I have proved that. But there is a difference between learning enough to build a product and learning enough to push a field forward. The second requires structure, mentors, and peers who are better than you at the thing you care about. A good university provides all three.
The application process itself is an exercise in strategic thinking. There are roughly twenty-five schools on my list, spread across the United States, the United Kingdom, the Netherlands, and possibly Switzerland. Each country has its own application platform, its own timeline, its own tests. The US wants the SAT. The UK wants the TMUA. Everyone wants IELTS. Some schools offer early action. Some offer restrictive early action, meaning you can only apply early to one of them. You have to decide, months in advance, which single school gets your best shot in November.
I am treating it like a project. Spreadsheets, deadlines, dependencies. Not so different from managing a product launch, except the product is the application itself and the customer is an admissions committee that reads thousands of them.
The schools that interest me most are the ones with serious ML research groups. Stanford, MIT, Carnegie Mellon, Edinburgh, the University of Amsterdam. Places where the people teaching the courses are also publishing the papers I have been reading. I want to sit in a room where the person at the front actually discovered the thing they are explaining, not just learned it from the same textbook I could read at home.
I know the acceptance rates. Stanford admits fewer than four per cent of applicants. MIT is similar. The Ivy League numbers are not encouraging for anyone, and they are particularly unkind to international students who also need financial aid.
But I have read enough admissions data to know that the numbers describe the pool, not the individual. Most applicants are strong students with good grades and solid extracurriculars. A smaller number have built real things, used by real people, generating real revenue, with real technical decisions behind them. A smaller number still have done that while founding and running a company. The pool is large. The overlap between "strong student" and "multinational company chairman at eighteen" is not.
I am not saying this guarantees anything. It does not. But it means the application is not a lottery. It is a case to be made, and I think I have a good one.
The part that sits least comfortably is the essay. Six hundred and fifty words to explain who you are and why you matter. Every guide I have read says to be authentic, to tell your real story, to avoid the temptation to inflate or perform. Fair enough. My real story is that I started building software during a pandemic lockdown at twelve, founded a corporation at fourteen, and have been running it since. The challenge is not finding material. It is deciding how much to reveal, and how to present the unusual without making it sound like showing off.
There is a fine line between confidence and arrogance in a university application, and the line moves depending on who is reading.
The honest reason I am doing this is not strategic. It is personal. I have spent the last six years building things. Products, teams, systems, a company. I am good at it. But I have reached the edges of what I can learn from building alone. The next version of what I want to build requires mathematics I do not yet have, theory I have only skimmed, and exposure to people who see the field from angles I have not considered.
I do not need a degree to keep running HMD Corporation. The company will operate regardless. But I want one, for the same reason I wanted to learn to code in the first place: because I ran into something I did not understand, and I wanted to understand it.
That instinct has not failed me yet.