A year inside the operating system
In first year of high school I spent a serious chunk of my free time trying to understand, from first principles, how Windows actually worked under the hood: processes, remote desktop, input handling, authentication. Some of what I built was a classic teenager-with-too-much-knowledge mistake; the technical foundation it left me with is something I still draw on every week.
What I was learning
- How Windows organises processes, sessions, and desktops, and why those distinctions matter.
- How remote-desktop protocols negotiate, frame, and stream input.
- How a process can read or restrict another process's input, and what the OS is doing under the hood when it allows or denies that.
- How auth flows actually work. Tokens, sessions, the boring middle that everyone forgets is where security lives.
The thing I'd undo
I built a remote-access tool and used it on a classmate's machine. I got caught. I deserved to get caught. There is a long distance between "I want to understand how this works" and "I have the right to run this on someone else's computer", and at fifteen I hadn't internalised it. I do now. I include this page so that distance is on the record, because "I was young and curious" doesn't excuse it, and pretending it didn't happen would be worse.
What stayed
A real working knowledge of how systems are put together, the kind you only get when you spend a year reading docs that were written for someone five years more senior than you are. It pays off constantly. When I write production ML or back-end services today, I think about auth surfaces, process boundaries, and the unhappy path because that year forced me to.
It also taught me, the hard way, that any capability needs a clear answer to "whose system, whose data, who agreed?". That answer comes first now.