AI Development Orchestrator

Location:

Remote Type:

Full-Time Reports to:

CEO Stack:

TypeScript, Python, React (Next.js App Router), Node, PostgreSQL, Supabase (Auth + RLS + Edge Functions + Realtime + Storage), Tailwind, shadcn/ui, Anthropic SDK, MCP, Bash, Git Stage: Revenue-generating SaaS, active production users Start:

Immediate

The Company Fabri Holdings is a dynamic and forward-thinking company that values innovative solutions and creativity. With a commitment to excellence and collaboration, the company provides comprehensive services tailored to the needs of its diverse clientele. Fabri Holdings fosters an inclusive and supportive work culture, encouraging employees to achieve their full potential. We are currently hiring for a partner company that is the operating system for founders building personal brands. They help B2B founders turn expertise into organic content engines that generate revenue without ads, agencies, or 10-hour content days. They have served 18,000+ founders and built a $15M/year business on organic content alone. They are now building the platform that powers our done-for-you delivery at scale.

The Role You are the engineering quality gate. You direct AI agents through focused task briefs, review every commit, and catch what the AI misses. You think in systems; the agents think in code. Multiple LLMs are run against the same problem; the strongest output is selected. Your job is to ensure what ships is correct, secure, performant, and architecturally sound. Every uncertain decision goes through structured review before approval. The validation layer is non-negotiable.

What You Do Review every AI-generated commit. Not \"does it run\" - \"what happens under concurrent load, what happens when an edge function cold-starts, what RLS policy is silently leaking across tenants.\" Architect features as single-task agent briefs: What / Why / Done When / Watch Out For. Scoped, focused, clear. Own the test pipeline. Vitest unit + Deno tests for edge functions + Playwright E2E. Evaluate whether tests prove behavior or inflate coverage. Diagnose, the agents fix, you verify. Refactor when AI-generated code drifts. Keep the codebase coherent. Own production stability. Read Supabase logs, trace edge functions through RLS, identify root cause, direct the fix. Scale architecture for 10x growth without rewrites.

Must-Haves Daily production AI workflow. You ship features where AI wrote the code and you validated it. Not autocomplete. Not casual. Multi-agent, multi-model, structured review. You are the reviewer; the AI is the author. Supabase in production. PostgreSQL with RLS, Auth (email + magic link + OAuth), Edge Functions (Deno), Realtime, Storage. You have been burned by race conditions and you know how to write RLS that is both secure AND performant. TypeScript primary. Strict mode. Zod at system boundaries. You understand why compile-time types and runtime validation are different problems. React in production. Next.js App Router, server components, complex state, real-time UIs, large component libraries (shadcn/ui or equivalent). You understand re-renders, memoization, and bundle size as engineering concerns. Python comfort. Our sync engines and integration scripts are Python. You can read, debug, and ship in it. Third-party API integration in production. Stripe, HubSpot, Calendly, Fathom, social APIs. Webhook verification, idempotency, retries, graceful degradation when an upstream is down. Edge function fluency. Deno or equivalent serverless. Cold starts, execution limits, no in-memory state across invocations. Code review depth. You catch security holes, race conditions, tests that pass without testing, RLS that leaks, cache that serves stale data after mutation. This is 60% of the role.

Strong Signals Built or customized AI agent workflows: Claude Code skills, CLAUDE.md configurations, memory persistence, multi-model comparison. Reviewed AI-generated PRs (not AI-assisted human code). Database-level access control in production. Real-time features: Supabase Realtime or WebSockets, with subscription lifecycle and reconnection handling. OWASP Top 10 fluency. Product SaaS background, not agency or consultancy.

Red Flags AI is autocomplete or boilerplate to you. Cannot explain architectural decisions in code shipped with AI assistance. Treats AI as infallible OR untrustworthy. Right stance: trust but verify. Overrides correct AI output on gut feel instead of validated review. Frontend-only or backend-only specialist. The review surface is full-stack. Consulting/body-shop background with no production ownership. No experience with real-time, data-heavy applications.

Who This Person Is 3 to 7 years full-stack, weighted toward TypeScript + Supabase (or equivalent PostgreSQL stack) and Python comfort. Not a 15-year veteran - we need someone agile and AI-native, not resistant. You understand a production system end-to-end: frontend, edge functions, database, auth, third-party integrations. You don't need to be the deepest expert in every layer. You need enough depth across all of them to catch when AI generates something architecturally wrong, and enough humility to validate before approving.

How We Evaluate No 60-minute coding test. We evaluate thinking, reviewing, directing. Bug Triage. Real (sanitized) bug from our codebase. Walk us through diagnosis. Do you read the code? Ask the right questions? Understand what could go wrong beyond the immediate symptom? Cold Code Review. Unaudited file written by an AI agent. Tell us what you see: security, race conditions, RLS leak, stale cache, tests that pass without proving anything. Task Brief. We hand you a feature requirement. Write the brief you'd give an AI agent. Is it clear, scoped, and likely to produce good first-pass output? AI Workflow Conversation. Describe your current AI workflow in detail. What you trust, what you double-check, how you handle uncertainty.

What We Offer A founding-level seat. Your judgment shapes the product. Direct CEO access. The most advanced AI development workflow you'll work in: multi-model agent orchestration, AI-generated codebase, structured validation, Claude Code skills + memory. Production impact day one. The product is live. Founders use it daily. Your work ships immediately. A product founders love: 18,000+ served, $15M/year, organic-only. A process that values getting it right. Speed comes from AI; quality comes from you.

To Apply or Refer Send a brief note describing your AI coding workflow and a link to something you've shipped. Include your React + TypeScript + Supabase (or equivalent) experience. No cover letter. We read code and we read people.