- 🧭 Frontend Chapter Lead at MasOrange (Madrid) — I help several product teams ship from one big Nx meta-monorepo: 5 independent workspaces, React 17→19, Next.js, Vite, Astro & friends, all the way to GKE
- 🤖 Currently all-in on AI-augmented engineering — specs, skills and agents baked into how we work every day, and getting more agentic by the sprint (more below 👇)
- 🥑 Trying hard to be a great teammate: mentoring, DX and docs are features too
- ⚡ Fun fact: some days our monorepo talks to more AI agents than humans 😄
AI isn't a side project for us — it's wired into the whole delivery loop, from the first idea to the merged PR.
And we say AI-augmented on purpose: the AI amplifies the engineer, never the other way around. As our loops grow more agentic — agents that plan, execute and verify their own work — exactly two things stay non-negotiably human: intent (what to build and why) and review (what actually ships).
| Stage | What we do | Powered by |
|---|---|---|
| 🧠 Shape | Spec-driven development: proposal → specs → design → tasks, humans and agents working from the same source of truth | OpenSpec + Jira |
| 🏗️ Build | Agent-ready codebase: AGENTS.md guides everywhere + 40+ reusable agent skills shared across our harnesses |
oh-my-pi · opencode · Claude Code |
| 🔌 Connect | Agents reach live docs, the browser, Slack and Jira through MCP servers | Model Context Protocol |
| 👀 Review | Team-aware AI code review on every PR — each team gets rules tuned to its own code | Claude in GitHub Actions |
| 📚 Learn | Skills as living knowledge: curated, versioned and shared — never locked into one vendor | skills-manager (my OSS) |
| 🔁 Evolve | Building loops that keep getting more efficient and self-verifiable — agents prove their own work before humans polish it | Us, always in the loop |
🧰 Peek inside the AI toolbox
- Spec workflow skills — propose, design, verify and archive changes without losing the paper trail
- Docs automation — TechDocs, ADRs, runbooks and catalog entries generated and kept honest by agents
- Guardrail skills — AI code review rules, writing de-slop, guided learning (
guide-me,quiz-me) so the humans keep growing too - skills-manager — my own open-source CLI to install, update and track agent skills across harnesses, privacy-first 🔒
💡 My take: AI is a force multiplier, not an autopilot — specs first, self-verifying loops, and intent & review are never delegated. 🧡
⚛️ Frontend at scale
🧰 Monorepo & tooling
🧪 Testing & quality
🚢 Platform & delivery
📡 Observability
Always happy to talk frontend at scale, AI-augmented workflows or self-verifying agent loops — say hi on LinkedIn 👋