Hillary Ibeanu

Berlin, Germany

Own the frontend for Feeds & Catalogs on Meta and other high-value channels (Pinterest, TikTok, Snapchat, Reddit, Google), powering product-catalog advertising for enterprise advertisers.

Platform & Product

• Implemented Pinterest and Reddit as new channels end-to-end, including catalog setup, feed field mapping, image templates and Pinterest multi-image templates; delivered as MDF lead, ahead of schedule
• Led the migration from component-based to config-driven channel support, cutting new-channel integration time by 28% and removing per-channel UI duplication
• Built LiquidJS-based feed mapping, improving feed processing success and page load speed
• Built the Catalog Activity Log and catalog health metrics, letting customers trace processing runs and self-diagnose failing feeds (~35% fewer related support requests)
• Led Segment event instrumentation for catalog and feed monitoring, following the tracking plan to measure adoption and funnel completion

Quality & Developer Experience

• Led Playwright E2E coverage across Meta, Pinterest, TikTok, Snapchat, Reddit and Automation Feed, covering full CRUD workflows and critical customer journeys; reduced regression-related customer issues by ~60–70% (Intercom conversations, CSM escalations)
• Led the migration from the legacy design system to accessibility-first components, and the modernisation of legacy class components into function components, with WCAG-aligned keyboard navigation and screen reader validation, enabling user-centric tests that query the UI by role and label and survive refactors
• Shifted the test strategy to user-centric integration tests, replacing module mocks with network-level mocking (MSW) and seeded, OpenAPI-typed fixtures, so tests exercise real components as users do; introduced Storybook on the same mocks
• Contributed to the Jest-to-Vitest migration in a pnpm and Turborepo monorepo: faster feedback loops, native ESM/TypeScript support and simpler test setup

Technical Leadership

• Led API contract design with backend teams (OpenAPI-first, generated typed clients), moving business logic from the frontend to the backend
• Led product reviews with design on feasibility, edge cases and empty, loading and error states
• Planned and pulled tech-debt work into sprints, balancing maintainability with feature delivery
• Mentored engineers through code review on composition over prop-driven APIs, self-contained components, state colocation, user-centric testing, server/UI state separation, URL-driven deep linking, Error Boundary + Suspense, layout-shift-free loading, optimistic updates, defensive programming and runtime API validation
• Drove AI-assisted development in the team: wrote agent rules and reusable skills (Cursor, Codex CLI) that encode our architecture, testing conventions and migration playbooks, used AI for first-pass code reviews, and connected Jira and Slack through MCP so agents work directly from tickets and threads, while keeping engineers accountable for review and verification

Germany

• Designed and built the international talent signup flow with React, TypeScript and Tailwind CSS, with automated tests in the CI/CD pipeline
• Ran A/B tests on signup variants using ABsmartly to improve conversion and user experience
• Contributed to job matching features, improving match accuracy between Talents and shifts
• Built an internal Retool tool (JavaScript + GraphQL) giving the Staffing Operations team visibility into shift fulfilment, Talent applications and bookings
• Contributed to backend microservices with NestJS (Node.js) and Redis
• Contributed to architectural decisions, technical refinements and code quality through code reviews

Portugal — Remote

Jungle.ai builds Canopy, an AI platform that detects abnormal behaviour in wind and solar assets from SCADA sensor data.

• Led the frontend redesign of Canopy V3, the flagship product, aligning stakeholders on requirements and refactoring the dashboard with accessible components for faster load times and better Core Web Vitals
• Built real-time alarm tracking with interactive D3.js visualisations, letting operators investigate context-aware alarms through time-series charts of sensor readings against model-predicted normal ranges
• Rebuilt the shared Vue component library used across all frontend projects, partnering with designers through design reviews to ensure consistency and reusability
• Introduced and led the adoption of Cypress end-to-end testing for the flagship product, catching regressions earlier and giving the team confidence to ship changes
• Contributed to user management features using Vue.js, TypeScript, GraphQL and Node.js, and maintained code quality through code reviews