Companies
Two companies built from scratch in under a year.
Holding company and IP vehicle that owns GHOBZ and any future ventures. Roukhos retains all intellectual property, architecture rights, and brand assets.
2 Active
Subsidiaries
LLC
Structure
About
The short version.
Moved to the US at 19
Left everything behind in January 2025. No connections, no safety net. Arrived with a laptop, a few ideas, and a clear sense that the only way to learn was to start building.
Moved to the US at 19
Left everything behind in January 2025. No connections, no safety net. Arrived with a laptop, a few ideas, and a clear sense that the only way to learn was to start building.
Self-taught everything
Never took a CS class. Every skill here was picked up by solving a real problem. Tutorials bore me. I learn by choosing a project just beyond my current level and figuring it out.
I think in systems
Before writing a line of code or making a business decision, I try to understand the underlying structure. What are the variables, what are the feedback loops, what breaks first under pressure.
Philosophy before strategy
Most decisions look complicated until you have a clear model of what you actually want. I spend more time getting the question right than finding the answer. Bad questions produce fast, confident, wrong conclusions.
Risk is a distribution, not a coin flip
I do not think about ideas as good or bad. I think about expected value across outcomes, and how much I can learn in the downside case. Most things worth doing have asymmetric payoff structures if you size them correctly.
Comfort with ambiguity
Immigration, payroll taxes, broken deployments, cold sales calls, all in the same week. Operating without a clear map is the default condition for anyone building something new. I have gotten comfortable with that.
Full-stack by necessity
When you are a solo founder you do not get to say that is not my job. I design, build, deploy, sell, and support. That range is a feature, not a compromise.
I test ideas cheaply and kill them honestly
Most ideas are not as good as they feel in the moment. The goal is not to believe in the idea. The goal is to find out if the idea is real as fast and cheaply as possible, then act on what you find.
Still just getting started
14 months in the US. Two companies. A growing stack of shipped products. And a clear sense that the most interesting problems are still ahead.
Moved to the US at 19
Left everything behind in January 2025. No connections, no safety net. Arrived with a laptop, a few ideas, and a clear sense that the only way to learn was to start building.
Self-taught everything
Never took a CS class. Every skill here was picked up by solving a real problem. Tutorials bore me. I learn by choosing a project just beyond my current level and figuring it out.
I think in systems
Before writing a line of code or making a business decision, I try to understand the underlying structure. What are the variables, what are the feedback loops, what breaks first under pressure.
Philosophy before strategy
Most decisions look complicated until you have a clear model of what you actually want. I spend more time getting the question right than finding the answer. Bad questions produce fast, confident, wrong conclusions.
Risk is a distribution, not a coin flip
I do not think about ideas as good or bad. I think about expected value across outcomes, and how much I can learn in the downside case. Most things worth doing have asymmetric payoff structures if you size them correctly.
Comfort with ambiguity
Immigration, payroll taxes, broken deployments, cold sales calls, all in the same week. Operating without a clear map is the default condition for anyone building something new. I have gotten comfortable with that.
Full-stack by necessity
When you are a solo founder you do not get to say that is not my job. I design, build, deploy, sell, and support. That range is a feature, not a compromise.
I test ideas cheaply and kill them honestly
Most ideas are not as good as they feel in the moment. The goal is not to believe in the idea. The goal is to find out if the idea is real as fast and cheaply as possible, then act on what you find.
Still just getting started
14 months in the US. Two companies. A growing stack of shipped products. And a clear sense that the most interesting problems are still ahead.
Feed
What I'm studying, building, and thinking about.
Shipped ARIA's per-client prompt architecture this week. The key insight: do not give clients a chatbot, give them a staff member that already knows their business. The prompt is not a template — it is a trained persona. Handoff detection was the hardest part to tune without false positives.
Built the skill map and timeline for this portfolio. D3's force layout is surprisingly intuitive once you stop fighting it. The trick is letting the simulation settle before freezing positions.
Implemented slide-in panels with AnimatePresence. Layout animations are magic for shared-element transitions but you need to be careful with exit animations. They can cause layout shifts if the container is not absolutely positioned.
Read through Patio11's pricing essays and applied the learnings to GHOBZ's Dominate tier. Key insight: anchor high, offer tiers, and never compete on price with someone who has more patience than you. The highest tier converted better than the lowest one did at launch.
Moved auth checks to Vercel Edge Middleware. Response times dropped from 200ms to 40ms for authenticated routes. Edge is the right layer for anything that does not need a full database round-trip.
Spent a week mapping what makes a small agency acquirable. Clean IP ownership, recurring revenue, documented SOPs, no single-founder dependency. Roukhos now owns all GHOBZ IP outright. A buyer gets the agency, not the systems behind it. That is the moat.
Built a semi-automated outreach pipeline: scrape leads, enrich with Clearbit, personalize with AI, send via Resend. 12% reply rate on the first batch. The key is genuine personalization — nobody responds to templates.
Hired a part-time operations person for GHOBZ. Learned more about employment law in two weeks than I expected. W-2 vs 1099 matters more than you think. Get a payroll service from day one.
A dashboard query went from 800ms to 12ms after adding composite indexes. EXPLAIN ANALYZE is your best friend. Most performance problems are missing indexes, not bad code.
Contact
Reach out — I'll get back to you.