Forward Deployed Engineer (AI/Automation)

We're a fast-growing ecommerce brand on Shopify with a serious AI and automation stack already in production. We're hiring a forward deployed engineer: someone embedded directly with our ops team who ships working software every single day, not someone who disappears for two weeks and comes back with a status update. You'll work alongside our existing AI engineer and report directly to our Head of Operations, with daily standups and a daily shipped-changelog. We have a large internal platform mid-build with ~9 interlinked projects. We move fast and we expect you to move faster. What's already running in production: - In-house AI CX agent (Python, FastAPI, Pydantic AI, multi-model LLM setup) handling support tickets end-to-end with Shopify integration - Internal creative platform (Next.js) with AI auto-tagging across - Meta Marketing API integration - n8n workflow automation, Langfuse observability, vector search What you'll build (with us, starting day one): - Internal tools and dashboards on our existing FastAPI/Postgres backend and Next.js frontends: creator ranking crons, brief-to-ad performance tracking, agency partner portal, product lifecycle dashboard, finance/performance analytics (Plaid, SKU-level profitability, ad spend vs inventory) - AI agents that read real business data and act: creative strategist agent surfacing spikes/correlations from ad + customer data, creator fatigue detection, trend detection from Reddit/social/post-purchase surveys - Pipelines wiring the Meta Marketing API (and later Google/TikTok) to our asset library so every creative is linked to its live performance metrics (ROAS, CTR, CVR, spend) - Automations across Shopify, Zendesk, Resend, Apify, and n8n: back-in-stock flows, review workflows, email sequences, scraping crons - Fast prototypes when we want to test an idea — same day, not next sprint — then hardened when they win You should be deeply familiar with: - Claude, GPT, Gemini — and which model to pick for which job - Agent frameworks: Pydantic AI, OpenAI Agents SDK, LangGraph, or custom orchestration - Claude Code, Cursor, or similar — we are an AI-first dev shop and expect you to leverage AI agents aggressively to multiply your output - FastAPI, Postgres, Redis, Railway, Cloudflare Workers/R2 - Next.js/React for internal tools (Clerk auth, clean role-based access) - RAG, embeddings, structured outputs, tool use, prompt caching - Ad platform APIs (Meta Marketing API especially) and analytics sources (Shopify, GA4, Triple Whale or similar) - Evals: Langfuse, Braintrust, Promptfoo, or custom harnesses - n8n / Zapier-class workflow automation - Shopify and Zendesk APIs - Webhooks, inbound email processing (Resend), scraping (Apify) How we work — read this before applying: - Daily standup with the Head of Ops. Daily end-of-day changelog in Slack listing exactly what you shipped. - PRDs live in Notion; you keep your tracker updated without being chased. - You ask for credentials/access when blocked instead of sitting on it for a day. - You propose better solutions when the spec is wrong. Pushback is welcome; silence is not. You're probably a fit if: - You read AI release notes the day they drop and have already tried the new thing - You can ship a working prototype in an afternoon and production-harden it the next day - You've built something where an agent actually reads metrics and decides what to do next - You think in systems, not scripts - You don't need hand-holding on infra, deploys, or debugging Start your proposal with the word "FORWARD" so we know you read this far.