Juan Obrach
I turn ambiguous product ideas into shipped systems. Built in scale-up teams, sharpened through independent product work across mobile, AI, backend, and infra.
Featured Case Studies
Evidence for technical leaders: product judgment, systems, tradeoffs, and shipped work.
Burger King Mobile App at Scale
Outcome Proof of production judgment in mobile ordering flows across USA, Europe, and Canada.
Mobile product work inside a real ordering business, where reliability, regional complexity, and team execution matter.
Flowtranslate AI Learning System
Outcome Turns daily translation moments into saved learning material, quota-aware AI work, and study paths.
A translation-first PWA that turns everyday translation history into personalized language learning.
Tablia Restaurant Menu AI Workflow
Outcome Compresses restaurant setup from messy menu input into review, publish, guest assistant, and analytics loops.
A B2B tool that turns messy restaurant menus into a live QR experience, an AI guest assistant, and owner analytics.
Latest Writing
Technical notes from building products, fixing systems, and working in the open.
claude-mem: Persistent Agent Memory Without Black Magic
A practical autopsy of how claude-mem works with Codex, MCP, hooks, local storage, and why enabled config does not always mean a plugin is loaded.
Read →Spec-Driven Development: Giving an Agent-Powered Repo a Real Memory
A practical guide to OpenSpec, GitHub Spec Kit, and how to organize durable product memory, agent workflows, and code in an AI-assisted monorepo.
Read →Maestro + Expo: The Real Gotchas Guide
Mobile E2E testing isn't hard. It's brutally environment-specific. A deep dive into every edge case we hit writing Maestro tests for an Expo app on iOS.
Read →