How a curious kid with a ZX Spectrum became an engineer who builds AI from scratch
When I was a kid in the early '90s, every boy in my class dreamed about the same thing: getting a game console or a ZX Spectrum to play games. And when they finally got one — that was it. Hours of Dizzy, Boulder Dash, and Elite. Pure entertainment.
I was different. Not because I was smarter or more disciplined — I just couldn't resist pressing the BREAK button.
The moment everything changed
I'd load a game from a cassette tape, wait through that awful screeching sound, and then — instead of playing — I'd hit BREAK. The screen would fill with cryptic BASIC code. And I felt something I can only describe as pure awe.
This is what's inside? This is how it works? These strange words and numbers — they make the game?
I was maybe 10 years old, and I was hooked.
The book that started it all
I found a book by Radionov and Larchenko — a Soviet-era programming manual for the ZX Spectrum. It was thick, technical, and absolutely not written for children. I didn't care. I read it cover to cover.
First came BASIC. Simple loops, PRINT statements, drawing pixels on the screen. But BASIC was slow, and I wanted more.
Then came assembly language.
I still remember the thrill of typing my favorite line:
ORG 40000
That was the starting address. From that point on, every byte was mine to control. No interpreter, no abstraction — just raw machine code talking directly to the Z80 processor.
My classmates were still playing games. I was writing them.
Building things nobody asked for
I didn't stop at simple programs. I got my hands on a Yamaha sound processor expansion board, connected it to my Spectrum, and wrote an assembly routine to digitize audio from cassette tapes. At an age when most kids were trading game cartridges, I was sampling sound and playing it back through code I wrote myself.
I created custom fonts — pixel by pixel, byte by byte, designing each character in a grid and encoding it into memory.
Then I wrote my own game. A small one, sure — but it had sprite animation, collision detection, and a game loop. All in Z80 assembly. All by hand.
And the feeling through all of this? It wasn't pride. It wasn't "look how smart I am for reading hard books at a young age." It was pure joy. The same joy I felt the first time I pressed BREAK. The joy of seeing how things work and making them do what I want.
From solder to software
That curiosity never went away — it just evolved.
From the ZX Spectrum, I moved to PC hardware — soldering, modding, understanding every chip on the motherboard. I fell in love with microcontrollers and circuits. I dove deep into C++ and low-level systems programming, always chasing that same feeling: what's inside? How does it really work?
This path led me to industrial automation — and it became my career for over 20 years.
20+ years in the field — and the fire never went out
I started at the bottom. Operator at oil field pumping stations in the Russian Arctic — the Nenets Autonomous Okrug, shift work in -40°C, maintaining instruments and control systems with my own hands.
Then I built an internet service provider from scratch. Got the licenses, installed base stations, deployed communication nodes at remote oil fields. I was the director, the engineer, and often the guy climbing the tower.
But the big chapter was Rosneft. Twelve years. I grew from an automation engineer to acting head of the production automation group — managing 105 people across 14 oil fields in the Komi Republic and Yamal. SCADA systems, telemetry, instrumentation, fire suppression, communication networks. I personally commissioned control systems at 3 new oil fields from the ground up.
I wrote software in C++, VBA, RAD Studio. I built monitoring dashboards in Wonderware InTouch. I designed databases in MySQL. When Flash was still a thing, I built interactive HMI panels in ActionScript. Whatever the problem needed — I learned it and built it.
After Rosneft, I moved to SCADA auditing and expertise — reviewing automation systems, writing technical specifications, quality control for one of the largest oil companies in the world.
And through all of it — every new controller, every protocol, every late night debugging a PLC — that same feeling from childhood was there. The joy of understanding how things work.
It never faded. Not once.
Why I'm building Genesis 2
Today, at 46, I'm building Genesis 2 — an AI system based on a novel Cascade Mixture-of-Experts architecture that I invented and patented. A system with 10,800 experts, 100% accuracy on its domains, 18ms inference — and it runs entirely on CPU. No GPU required.
It doesn't just generate text — it executes. It runs code, manages servers, solves real engineering problems. Because that's what I've always built: things that do things.
People ask me: "Why build your own AI from scratch? Why not just use APIs?"
Because I'm still that kid pressing BREAK.
I don't want to use the black box. I want to open it, understand every neuron, every weight, every routing decision. I want to build something that works the way I think it should work — efficient, practical, and deeply understood.
The ORG 40000 of today is a neural network's first layer. The Yamaha sound chip is now a GPU. The Z80 opcodes are now PyTorch tensors. But the feeling? The feeling is exactly the same.
I love this. I loved it when I was 10, I love it at 46, and I'll love it at 80. It's not a job. It's not a career move. It's the way my brain is wired — and I wouldn't change a thing.
The point
I'm not writing this to brag. I'm writing this because I think the world needs more people who press BREAK.
Not people who are afraid of complexity. Not people who need permission to learn something "too advanced." Just curious people who look at a running system and think: I wonder what's inside.
If you're that person — whether you're 10 or 50 — trust that instinct. It's the most valuable thing you have.
I'm Aleksandr Larionov — industrial automation engineer with 20+ years of experience, inventor, and builder of AI systems. Find my work on GitHub.
Top comments (0)