• You build and operate AI-powered end-to-end workflows — from ad creative production through content orchestration to GTM automations — ensuring the entire team benefits. • You develop a central orchestration layer based on Claude Code and MCP that connects different skills and tools without requiring team members to understand the underlying architecture. • You integrate our core platforms — HubSpot, Asana, Meta Ads, Google Ads, Clay and others — directly into automated workflows and ensure clean, reliable data flows between systems. • You establish structured feedback loops so that performance data continuously feeds back into foundation files (brand voice, ICP, positioning) and the system improves with each iteration. • You take ownership of concrete deliverables from day one: within the first 90 days you will deploy a complete ad creative workflow and a content orchestrator that the team can use without instruction. • You create a structured roadmap for the team’s existing AI approaches — identifying what can scale, what doesn’t work, and what should be built next.

• Planung, Architektur und Entwicklung von Shopify Apps, Plugins und API-Integrationen • Full-Stack-Entwicklung mit modernen Technologien im Frontend (React, Next.js) und Backend (Node.js, Express, TypeScript) • Anbindung von Drittanbieter-Systemen über REST- und GraphQL-Schnittstellen • Sicherstellung von Performance, Code-Qualität und Skalierbarkeit • Umsetzung responsiver und zugänglicher UI-Komponenten • Durchführung von Code Reviews, Testing (Unit, Integration, E2E) und Debugging • Aufbau und Pflege von CI/CD-Pipelines sowie technische Dokumentation • Zusammenarbeit mit Design, Product und QA, um Produkte von Grund auf zu gestalten

• You develop and optimize Shopify apps • You turn ideas and concepts into code • You create scalable infrastructure • You build simple apps & layouts

• Own agent quality end-to-end: diagnosis, improvement, and validation across SmartAssist's orchestrator and subagents • Identify failure modes across quality dimensions factual accuracy, completeness, tone, actionability, and latency and prioritize what to fix • Drive quality improvements through prompt engineering, context engineering, and RAG retrieval tuning • Extend and mature our evaluation framework: scorers, golden datasets, regression gates, and online evaluation for production traffic • Close the feedback loop ensure that every change has a measurable, attributable quality signal • Collaborate with our Agent Architecture lead to distinguish quality problems that require prompt/context solutions from those that require structural fixes • Establish repeatable methodology that scales beyond any single agent or subagent