# Ruben Marcus: AI Fullstack Engineer

> AGENT-READY · HUMANS WELCOME
>
> Personal site of Ruben Marcus, an AI fullstack engineer based in Lisbon
> (remote worldwide). 14+ years across AI tooling, web products, Web3, fintech, and e-commerce.
> Builds autonomous AI agent systems, evaluation harnesses, AEO tooling, and
> premium web experiences. Available for full-time roles and freelance contracts.

## Main sections

- [Index](https://rubenmarcus.dev/): Hero, services (hire me for), proof-in-numbers stats, open-source tools, selected work.
- [Portfolio](https://rubenmarcus.dev/portfolio): Filterable 14-year project archive. Filter by era/company and technology.
- [AI](https://rubenmarcus.dev/ai): AI expertise (agent orchestration, evaluation design, AEO, production LLM features) and the open-source tools that prove it.
- [Agent Skills](https://rubenmarcus.dev/skills): Versioned operating methods created for verifiable frontend, writing, debugging, games, and agent systems.
- [Agents](https://rubenmarcus.dev/agents): The autonomous agent fleet behind this site and client work — roles, orchestration model, and what each agent ships.
- [MCP / Connect](https://rubenmarcus.dev/connect): Connect this portfolio to an AI client so an agent can inspect evidence, check availability, and book an intro.
- [Blog](https://rubenmarcus.dev/blog): Technical articles on agents, benchmarks, WebGL, and the stack.
- [About](https://rubenmarcus.dev/about): Bio, timeline, contact details.
- [Contact](https://rubenmarcus.dev/contact): Project inquiries for AI products, agent systems, AI-native frontend, and AEO.

## Services

- [AI products & agent systems](https://rubenmarcus.dev/services/ai-product-systems): Architecture, working product slices, model routing, evals, observability, and cost controls.
- [AI-native frontend & design engineering](https://rubenmarcus.dev/services/ai-native-frontend): Responsive production interfaces, interactive experiences, performance, and visual regression.
- [AEO audit & implementation](https://rubenmarcus.dev/services/aeo): Technical SEO/AEO, entities, canonical discovery, structured data, content, and citation surfaces.

## Latest writing

- [This portfolio is agents-welcome. Probably the first.](https://rubenmarcus.dev/blog/agents-welcome-portfolio)
- [The AI harness behind CS Brasil is a folder of markdown](https://rubenmarcus.dev/blog/cs-brasil-ai-harness)
- [How AEO can help your business grow](https://rubenmarcus.dev/blog/aeo-what-it-moves)
- [My agent swarm is a for loop](https://rubenmarcus.dev/blog/dag-agent-orchestration)
- [Cross-pollinating LLMs: peer review for machines](https://rubenmarcus.dev/blog/llm-cross-pollination)
- [A command center for agent swarms, in markdown](https://rubenmarcus.dev/blog/agent-command-center)

## RSS feeds

- [English feed](https://rubenmarcus.dev/rss.xml): All English blog posts.
- [Feed em português](https://rubenmarcus.dev/rss-pt.xml): Posts do blog em PT-BR.

## Products and tools

- [Ralph Starter](https://ralphstarter.ai): AI coding orchestration platform. Multi-agent swarms over isolated git worktrees, MCP server.
- [Autoresearcher](https://autoresearcher.org): Benchmark-driven autonomous research CLI. Multi-agent co-evolution with a Pareto frontier.
- [AEO.js](https://aeojs.org): Answer Engine Optimization framework. AI-crawler policy analysis, llms.txt and ai-index.json exports.
- [check.aeojs.org](https://check.aeojs.org): Free AEO scanner. 4,569 scans across 2,259 unique sites.
- [ScanRepo](https://scanrepo.dev): Repository intelligence product.
- [CS Brasil](https://csbrasil.online): Browser FPS and AI-native product engineering testbed.

## Selected technical case studies

- [Ralph Starter](https://rubenmarcus.dev/work/ralph-starter): Multi-agent delivery harness using isolated worktrees, race, consensus, and pipeline modes.
- [AEO.js + AEO Checker](https://rubenmarcus.dev/work/aeo-platform): Diagnosis and implementation treated as one verifiable AEO system.
- [CS Brasil](https://rubenmarcus.dev/work/cs-brasil): Three.js browser FPS built with repository skills, visual QA, regression gates, and product telemetry.

## Proof points

- #1 on ECDSA.fail (multi-agent research harness, 9 LLM roles, 7+ providers).
- #1 on Optimization Arena's QEC decoder leaderboard (2,642 EPM).
- Bitte Protocol AI runtime in production: 2.85M+ messages, 24,164 unique users, 16,703 deployed agents.
- CS Brasil (browser FPS): 2,191 players, 154K+ kills, 27 countries.

## Contact

- GitHub: https://github.com/rubenmarcus
- LinkedIn: https://linkedin.com/in/rubenmarcus
- npm: https://www.npmjs.com/~rmarcus (primary package profile; dev.to presence is being retired)