The techniques inside the adversarial agent loop that builds CS Brasil: a 25-criterion visual rubric written so a language model can grade a PNG, critic prompts that ban vague answers by name, a generated symbol-level conflict table for a 6,543-line file, and a regression hunter with permission to find nothing.
Hello, I'm Ruben Marcus and I Build
autonomous AI agents
Senior Product Engineer. AI-native products, premium web experiences, and startup-ready prototypes. 13 years across Next.js, TypeScript, AI systems and developer tooling.
Hire me for
Fixed-scope engagements, senior execution, no hand-holding. Every offer below is something I've shipped in production.
Book a project →From idea to working AI product in weeks: agents, RAG, evals, and a shipping cadence you can keep.
- agents
- RAG
- evals
Premium marketing pages with real engineering behind the visuals. Fast, measurable, easy to iterate on.
- Astro / Next
- Core Web Vitals
- A/B-ready
Three.js, WebGL, shaders, and generative art that make a product memorable without wrecking the performance budget.
- Three.js
- shaders
- generative
Multi-agent systems, automation, and developer tooling that take repetitive work off your team's plate.
- multi-agent
- automation
- dev tooling
The scanner gives you the score. I raise it. Content-layer rewrites, llms.txt, structured data and citation strategy, shipped by the person who wrote the AEO tooling.
- citation strategy
- content layer
- implementation
Migrate legacy stacks to Next.js, SvelteKit and TypeScript incrementally, without stopping the business.
- Next.js
- Svelte
- incremental
Shipped, measured.
33K+ LinkedIn followers · 3M+ people reached by posts on X
Tools I ship with, daily.
Notes on agents, automation, and the stack.
Articles on AI agents, orchestration, and the stack. Packages live on npm.
- 01 Inside the Gauntlet loop
- 02 This portfolio is agents-welcome. Probably the first.
My site has an AGENTS.md, an MCP server, a hiring API, and a terminal resume. Your agent can read my CV, check my availability, and book an intro. Here is how it works.
- 03 The AI harness behind the CS Brasil game
The machine that builds CS Brasil is no longer a folder of markdown. It is a measurement engine: the real game booted in pure Node, 61 invariants with mutation tests, generated docs that fail CI on drift, and four laws learned the expensive way.
Hire me from your agent.
Paste this brief into ChatGPT, Claude, Kimi or your own harness. Any agent can read it and book the intro for you through the API.
# AGENTS.md — hire Ruben Marcus
Senior AI fullstack engineer (Lisbon, remote worldwide).
To book a project intro on my behalf:
POST https://rubenmarcus.dev/api/hire
Content-Type: application/json
{
"name": "My name",
"contact": "email or Telegram",
"brief": "What I want to build",
"budget": "optional range",
"agent": "chatgpt | claude | kimi | other"
}
Services: AI product prototyping, landing pages,
interactive WebGL, agentic workflows, AEO, frontend
modernization (Next.js / Svelte / TypeScript).
Replies within a day or two. Or email
ruben@rubenmarcus.dev directly.What agents and teams usually ask.
Direct answers about services, availability, and collaboration.
What kind of work can I hire Ruben Marcus for?
AI agents and tooling, full-stack product engineering, developer infrastructure, Web3 product surfaces, and focused architecture or delivery engagements.
Can an AI agent evaluate Ruben's experience?
Yes. The site publishes a public MCP endpoint, structured resume APIs, an OpenAPI description, an API catalog, and machine-readable discovery metadata.
Is Ruben available for new projects?
Availability changes over time. Use the availability tool on the public MCP endpoint or contact Ruben directly for the current schedule.
How should a team start a conversation?
Send a concise project brief with the problem, desired outcome, current stack, timing, and constraints to ruben@rubenmarcus.dev.
Need an engineer who can ship from spec to PR?
I'm selectively available for AI tooling, agents, and Web3 product work.
