LOG // CLOUDFLARE // MARTECH // AI_PRODUCTION

Former Senior AI-Driven Web Producer

Cloudflare: MarTech Engineering & Agentic Web Production (Completed Engagement)

Pioneered agentic AI systems and automated workflows to build and deploy complex, accessible content types in a fraction of the time. Bridged front-end engineering, experimentation, and design system governance for global digital growth. Now: Mentor & Lead Architect, AgentLionz Co.

Key Responsibilities

AI Bespoke Page Builder

Built a custom AI agentic skill using OpenCode that constructs complex, brand-compliant Contentful pages in 5 to 10 minutes — a task that previously took 1 to 2 days of manual work.

Document-to-Page Agent

Empowered Product Marketing Managers and stakeholders to upload raw Google Docs or Figma designs. The custom agent instantly generates dynamic, component-driven pages using pre-built engineering fragments.

Strict Automated QA Rules

Set up strict AI validation rules within the agent to automatically resolve QA issues — including camelCase enforcement, image sizing, and brand compliance — eliminating manual back-and-forth between teams.

AI-Driven Content Migration

Extended the agentic system to a large-scale Contentful migration to the site's new platform — agents fetched legacy page content and populated newly created content types, replacing a previously manual migration process.

Adobe Target A/B Testing

Engineered interactive components, landing pages, and A/B test variants (HTML/CSS/JS) via Adobe Target to optimize conversion funnels.

Technologies

OpenCode · Windsurf · Claude · Cloudflare Workers AI · Contentful · Astro · HTML / CSS / JS · Adobe Target · Figma

PIPELINE // ARCHITECTURE // MCP

From Figma file to live page, without the handoff

An agentic system connected to Contentful via MCP. It reads the design and the documentation, builds the page, and checks its own work against the design system before a human ever needs to look at it.

Input Figma + Docs Approved design files and brand / design-system documentation.

Agentic system · MCP AI Agent Reads the design, generates the full page, QA's the output automatically.

readsgeneratesself-QA

CMS Contentful Structured entries across 74+ distinct content types.

Gate Automated QA Every output checked against brand guidelines and the design system.

Output Live Page Published on cloudflare.com — minutes after the brief, not days.

Same system, second job — the replatform migration

Legacy Contentful pages Agents fetch content data New content types, created & populated New site version

A large, complex migration that had been handled manually — automated end-to-end, including schema-level content-type changes.

BEFORE // AFTER

The same page, two ways

What producing one marketing page looked like manually, next to what it looks like through the pipeline. Same brief, same quality bar — the difference is who does the repetitive work.

Manual production ~2 days

Brief & design handoff Figma file reviewed, requirements interpreted by hand.

Manual page build Entries assembled field-by-field in Contentful.

Manual QA Checked against brand guidelines by eye — the build/QA step alone took ~4 hours.

Fix rounds & publish Back-and-forth until it matches the design.

Agentic pipeline ~5 minutes

Brief Point the agent at the Figma file and the docs.

Page generatedauto Full page built directly into Contentful via MCP.

Self-QAauto Output checked against brand guidelines and the design system — build/QA now ~30 minutes when run as a standalone step.

Human review & publish A person approves; the machine did the repetition.

87%

Build/QA step reduction — 4 hours to 30 minutes

74+

Distinct content types under AI agent deployment

Days → Minutes

End-to-end page turnaround, automated

CONTINUE // EXPLORING

What's Next?

Explore other engineering case studies, or dive into my interactive career story.

Explore Other Topics //

Toyota Experience Enterprise CMS migration at 41-country scale, the manual way
view case study →
William Hill Experience Large-scale A/B testing and React components for millions of users
view case study →
💡 Pro Tip Ask Agent Lionz: "How did Pedro's Cloudflare pipeline QA pages automatically?"

← Back to Projects