avatar

Jorge Zozaya

Full Stack Engineer | ML Innovator
Dynamic Peace

Soy un ingeniero full-stack que construye la parte de la IA que la gente realmente conoce: agentes seguros para la marca, frente al cliente, y los sistemas que los mantienen honestos. La parte difícil no es hacer que un modelo responda — es hacer que responda como quien firmó un acuerdo legal. Mi trabajo reciente: un chatbot de WhatsApp diseñado para sobrevivir cinco rondas de revisión con el equipo legal de Nike, donde cada emoji, cada interjección y cada válvula de escape conversacional fue auditado, y una máquina de estados — no un prompt — imponiendo las reglas.

Experiencia y habilidades

  • Montalvo Wire · Since 2021

    Powering one of the biggest names in the advertising world in Mexico.

    AlgosUX/UIBackendAI
  • Clay GTM Case Study · Sep 2026

    $0 platform credits — 56 production houses scored. Written in public, in progress.

    GTMClayRevOpsLead Scoring
  • Brand-safe LLM chatbot on WhatsApp — five persona revisions, a 7-state machine, server-enforced guardrails.

    LLMWhatsAppTwilioPythonSupabase
  • MoveNet pose estimation turned dance into shareable content.

    TensorFlowMoveNetPose Estimation
  • Commented Mar 2024

    "Jorge met and exceeded expectations and became a truly valuable team member on our project. Without his help, we would have been hard-pressed to meet deadlines, and his work was of the highest standard."

  • Magnum Mini OCR · Nov 2023

    Receipt-scanning OCR web app for Unilever — tickets turned into rewards.

    OCRComputer VisionNext.js
  • ML fan experience that brought Mexico closer to the game.

    Computer VisionMLNext.js
  • Commented Jun 2022

    "It was a pleasure working with Jorge. He is very professional in React.js. He also understands the tasks well, presents various solutions, explains their advantages/drawbacks and then executes a methodical and efficient resolution. I would highly recommend Jorge without hesitation. A+++5++++5++++5++++5++++!"

  • Headless storefront for Moleskine MX — live at moleskine.com.mx.

    ShopifyHeadlessSanity CMS

La otra mitad de mi práctica es la ingeniería de GTM: probar el mercado antes de invertir en él. Califiqué 56 productoras y gasté $0 en créditos de plataforma.

Tres tarjetas de evidencia: 10+ años entregando; 5 revisiones de persona, 1 voz que sobrevivió; $0 en créditos gastados calificando 56 productoras

El caso de estudio GTM de $0 en créditos se está escribiendo en público ahora mismo.