What you'll actually build

• Our client + internal portals (Next.js + Supabase). Real users, real product surface — your features ship the same week you build them.
• 200+ live n8n workflows powering CRM, lead enrichment, billing, system health, and YouTube ops. You own this stack — debug it, extend it, replace the fragile parts.
• Our AI video production pipeline — fal.ai, Kling, ElevenLabs, Remotion, FFmpeg. The long-term goal: creatives contribute only the idea, the pipeline handles prompting → voiceover → footage → cuts → subtitles → render → delivery automatically. Multi-year arc, real impact.
• The integration layer that ties everything together — Close CRM, Meta Ads API, Airtable, SevDesk, Google Workspace.
• Our infra — Hetzner VPS today (Docker, Traefik, PM2, 15+ microservices), migrating progressively to Hostinger + Vercel for portals.

Why work here

• You ARE the technical backbone — not "a dev on a team". No JIRA committee. No design-review purgatory. You spec → build → ship.
• Direct access to the founder. Decisions in minutes, not in two sprints. No middle management.
• AI-first by default. Claude Code (Opus + Sonnet) and Codex are funded and expected — the way we ship faster than traditional teams. If you already live in Claude Code, you'll feel at home day one.
• Modern stack only. No legacy Java / PHP. TypeScript, Next.js, Supabase, Python, n8n, Docker.
• Real ownership, real upside. Performance bonus tied to automations that replace manual ops — we measure hours saved, you get paid for it. Voluntary 13th-month bonus on top.
• 100% remote, async-first. Loom + written specs over sync meetings. Minimum 4h/day CET overlap, otherwise your hours are yours.
• Mission you can be proud of. We're automating an entire creative production pipeline end-to-end. The code you ship touches thousands of videos a year.

Tools you'll use daily

What we're looking for in you

• Diagnostic intuition (most important): Roughly 80% of the day-to-day is reading and fixing unfamiliar code. You can pick up an n8n workflow or Supabase schema you've never seen, read it cold, and form a working hypothesis about what's broken within minutes.
• AI-skepsis & full ownership: You ship daily with Claude Code / Codex as a multiplier — but you know where the AI hallucinates and you catch it before it ships. You can defend every line in a live screen-share, AI-generated or not. "The AI wrote it" is not an answer in our team.
• Real n8n experience: Production n8n, not tutorials. You know $json vs $node["X"].json vs $items(), and when logic belongs in n8n vs. a custom microservice.
• Strong backend fundamentals: Supabase / Postgres (RLS, Auth, indexes, unique constraints, ON CONFLICT), REST APIs, TypeScript or Python (one solid, not both half-decent). Solid SQL.
• Operational discipline: You follow SOPs, update the knowledgebase when you change things, and test before deploying — without anyone reminding you.
• Linux / Docker comfort: SSH, read logs, restart services, debug a failing container.
• Strong written English: All specs, PRs, post-mortems, and internal comms are in English.
• Async discipline: Loom walkthroughs, written specs, Mattermost-first. No daily sync standups.

Bonus skills (not required)

Who this role is NOT for

• Code purists who refuse AI-assisted development
• Blind AI pasters who ship code they can't explain line by line
• "It works on my machine" developers — every deploy gets an E2E test
• People who skip docs — modifying an n8n workflow without checking the knowledgebase breaks downstream ops
• People who need constant direction — this role is autonomous
• Silent workers — no async update for a full day = red flag
• Framework fanatics — we pick pragmatic tools, not the framework of the month

About realistisch.ai

How to apply