In 2016 I didn't know what a server was.
I was a kid uploading ARK: Survival Evolved videos to YouTube, and the game — as shipped — was unrecordable. Taming took days. Walking took hours. So I started digging into config files: player speed, taming multipliers, breeding timers. Trial and error, on my own machine, until it was recordable. I didn't know it yet, but that was my first product decision: the game had to fit the format, so I changed the game.
The channel grew. I got invited to multiplayer series with bigger creators. And one day, mid-series, the guy running the server had a falling-out with the owner and walked. Nobody else knew how to keep it alive. I said: I'll run it. I had never run a real server. Three days later the series ran on it. I've said that sentence for every job I've taken since.
The conversation that changed how I quote work
Word spread that I could handle ARK servers. Then Willyrex, a top Spanish-speaking creator, joined a series I was running and asked if I could set up Conan Exiles and GTA V servers too. I had no idea how. I said yes, learned, and delivered.
But that's not why he kept coming back. He told me why: every technical person he'd worked with said yes to everything and then blew the deadline. I did the opposite. When he asked for something enormous, I'd say: "What you actually want for the video is THIS, right? Building it exactly as you describe is madness — weeks of work. But framed this way, you get almost the same result on screen, and you'll have it tomorrow."
He and Vegetta777 loved it. I didn't have a word for it then. Now I do.
That was the start of ten years of commissions: an ARK series with Pokémon-style gyms and creatures (with Staxx in it), servers for Rubius, xFarganx, and others. They came to me with a game and a theme; I owned everything else — research, team, budget, game systems, and enough fresh content that each creator had material every single episode. They kept commissioning me across different games, because what traveled was the execution, not the game.
Shipping on a platform nobody had documented yet
When Epic launched Fortnite Creative, Vegetta had a sponsored campaign with them — he and Fargan needed maps to showcase the new mode. The mode was less than a month old. No Verse, no documentation, no best practices, nothing to copy. I called three friends — Arsilex, Jesusseron, Hernybreak — and we built our first two maps under constraints nobody had mapped yet. The maps shipped and the campaign ran.
Some time later, Vegetta asked me how I'd approach a series around creator-made maps. I told him two things were missing: more people on screen, for dynamics — and maps built expressly for the group, not generic ones. I took care of both. By mid-2019 that group of friends had a name the audience gave us — Los Noobs — and a series on Vegetta's channel, Minecraft con Noobs, alongside Fargan, Elyas360 and Alexby. Vanilla Minecraft, me on camera as a participant, at a time when nobody was making Spanish-speaking Minecraft series.
Running my own platform, on my own payroll
Then I built Tomatecraft, my own Minecraft network. The product decision that made it work: Java (PC) and mobile players could finally play together, when almost no server offered crossplay. I funded it myself, out of a channel that by then had passed half a million subscribers, with programmers, designers and builders on payroll. Later I merged networks with Arsilex and we ran large branded events — up to 20 contributors per event, with Vegetta, Willyrex, Fargan and Rubius joining in.
Every feature I cut, I cut knowing what the payroll cost that month.
Then I stopped for a long while — health comes first. When I came back, I formalized what I'd been doing by instinct and earned a Java certification.
The comeback: 65,000 concurrent players
Years earlier, when Epic ran the Fortnite World Cup and Rubius needed an official map for his qualifier — RubiusTrials — I had pushed hard for two builders I believed in: Hooshen and Iscariote. I didn't build a single brick of it; I just insisted, to Vegetta and Rubius, that they were the right ones. They were.
So when those same two needed someone to lead projects and bring in clients, they called me. I directed the Fall Guys collaboration island (with Vegetta777 and Willyrex): 65,000 concurrent players at peak, the #1 most-played island globally across all of Fortnite for several hours on day 2, and #1 in its category for two weeks — ahead of Epic's own official maps for the event.
Afterwards, with Los Noobs, we shipped island after island — I directed five or six of them end to end, with a core team of five, and contributed to twenty-plus more in specialist roles: gameplay programming, configuration, 3D modeling, mechanics design, and pricing. We monetized through Epic's engagement payouts, and when in-island micro-purchases arrived, pricing became my most common role across those twenty-plus islands. My last project with the studio was a battle-pass system: VIP pass, level purchases, score multipliers — built as a reusable framework, adopted across 4 maps by the time I left. What I learned pricing them: the hardest purchase is the first one. After that, buying again feels natural — so I priced the first one to be bought.
Today: AI agent systems
Today I build AI agent systems. My 19-agent orchestration went from advisory rules — followed 0 times out of 3 — to deny-by-default enforcement: across 7 runs, 9 blocked tool-calls, 4 allowed only via explicit logged override, 0 unauthorized writes — published with reproducible numbers, extracted into an open-source package (loopward). Different stack. Same rule: the deadline is part of the product.
I never held the PM title. I have the proof.
