Org-wide AI code review agents
Engineered the context, prompts and rules behind agents that review frontend merge requests and autofix accessibility issues — integrated with Jira and GitLab, hardened against prompt-trust and token-leak gaps.
I've spent 15 years building React platforms and design systems. Now I spend most days teaching AI agents to build them well.
I orchestrate Claude Code, Cursor and MCP to ship full-stack features faster, and write the context, rules and skills that keep agent output production-grade. A philosophy degree taught me the part that matters most: say exactly what you mean, then check the work.
Engineered the context, prompts and rules behind agents that review frontend merge requests and autofix accessibility issues — integrated with Jira and GitLab, hardened against prompt-trust and token-leak gaps.
About ten reusable skills that automate the dev loop: ticket kickoff, QA test plans, staging deploys, test-data seeding and Storybook verification.
Led the frontend for a greenfield React SPA in a new pnpm/Turbo monorepo — magic-link auth, time-zone-safe scheduling, English/Spanish — shipped ahead of the backend against stubs.
An MUI-based component library with metadata-driven fields that significantly cut UI build time for new social-services features.