AI Development Orchestrator (Remote)

Automation Expert - Architect Freedom Through Systems That Scale

Location: Remote
‍Type: Full-Time
‍Reports to: CEO / Head of Product
‍Stack: React 19 · Vite · TypeScript · Supabase · Tailwind · shadcn/ui
‍Stage: Revenue-generating SaaS, active production users
‍Start: Immediate

The Company

Founder OS is the operating system for founders building personal brands. We help B2B founders turn their expertise into organic content machines that generate revenue — without ads, without agencies, without spending 10 hours a week on content. We have served 18,000+ founders, built a $15M/year business on organic content alone, and are now building the technology platform that powers our done-for-you delivery at scale.

The platform is a production SaaS application used by our internal team and founders daily. It manages the entire personal brand lifecycle: content generation, brand strategy, asset delivery, publishing workflows, calendar management, lead magnet creation, and founder engagement. It is integrated with Supabase for auth, database, and edge functions, and designed for scale.

We are an AI-first company. The CEO builds products directly using AI coding agents. The engineering culture is fast, opinionated, and focused on shipping an outstanding product that founders love.

The Role

We build with AI. The codebase — frontend, backend, edge functions, infrastructure — is primarily authored by AI coding agents (Claude Code, Cursor, Codex) orchestrated by the product team. The person in this role is the engineering quality gate. They review, architect, test, refactor, and validate— ensuring what ships is production-grade. They direct AI agents through focused task briefs and catch what the AI misses.

This is an AI-native development operation. Multiple LLMs are run against the same problem, outputs are compared, and the strongest solution is selected. The orchestrator ensures that what ships is correct, secure, performant, and architecturally sound. They direct the AI agents through focused task briefs, validate every commit against production standards, and run uncertain decisions through a structured review process before approving.

The validation layer is non-negotiable. You always ask. If you understand the big picture, you ask better questions — but you never skip the step. The process catches what individuals miss.

What will you do

• Review every AI-generated commit for correctness, security, performance, and architectural consistency. Not 'does it run' — 'what happens when two requests hit this at the same time' and' what happens when Supabase Edge Functions cold-start under load.'
• Architect features as single-task briefs that AI agents execute against. Format: What / Why /Done When / Watch Out For. No monolithic specs. Focused, scoped, clear. You think in systems; the agents think in code.
• Own the testing pipeline. Ensure AI-generated tests cover meaningful scenarios. Own the Playwright E2E suite and Vitest unit test coverage. Evaluate whether test suites actually test real user behavior, not just inflate coverage numbers.
• Identify and direct refactoring when AI-generated code accumulates technical debt, inconsistent patterns, or performance bottlenecks. You diagnose, the agents fix; you verify. The codebase stays clean.
• Own production stability. When something breaks, you are the person who reads the Supabase logs, traces the issue through edge functions and RLS policies, identifies the root cause, and directs the fix. You have been on-call before, and you know what a production incident feels like.
• Enforce the validation process. Every uncertain decision goes through a structured review. Every commit is reviewed. Every deployment is tested. No shortcuts, no cowboying features into production. We ship fast, but we ship right.
• Scale the platform architecture. The product serves hundreds of active founders with real-time features, content generation workflows, and complex data relationships. You ensure the architecture can handle 10x growth without rewriting.

Must-haves

Frontend & Full-Stack

• Deep React experience in production. Not tutorials — production applications with complex state management, real-time updates, and data-heavy UIs. You understand when to use React Query for server state vs. local state, how to structure a large component library, and why re-renders matter.
• TypeScript is your primary language. You write strict TypeScript. You use Zod for runtime validation at system boundaries. You understand the difference between type safety at compile time and validation at runtime, and why you need both.
• Experience with Vite, React Router, and modern React patterns. Server components aren't the only paradigm. You understand SPA architecture, client-side routing, code splitting, and how to build performant single-page applications that feel fast.
• Component library experience. shadcn/ui, Radix UI primitives, or equivalent. You understand compound components, accessibility patterns, and how to build a consistent design system that scales across hundreds of components.

Backend & Infrastructure

• Has built and shipped with Supabase in production. Not just used it — built real features on it. PostgreSQL with Row Level Security policies, Supabase Auth (email, magic link, OAuth), Edge Functions (Deno runtime), Realtime subscriptions, and storage. Knows what breaks and why.
• Understands database design, RLS policies, and race conditions from experience. Has been burned by them. Can explain what happens when two concurrent requests try to update the same founder's content calendar simultaneously. Knows how to write RLS policies that are secure AND performant.
• Has dealt with auth and session security. Supabase Auth, JWT validation, refresh token flows, and session management. Knows why storing tokens in localStorage is a footgun and how to handle auth state across tabs and page reloads.
• Has integrated third-party APIs in production. Stripe, ManyChat, Fathom, HubSpot, social media APIs, or equivalent. Knows what webhook verification is and why idempotency matters. Has built resilient integrations that handle API downtime gracefully.
• Has experience with edge function deployment. Supabase Edge Functions (Deno), Vercel serverless, AWS Lambda, or equivalent. Understands cold starts, execution limits, and why in-memory state does not work across invocations.

Code Review & Architecture

• Can read someone else's code critically. Specifically, AI-generated code. Catches security holes, performance anti-patterns, incorrect assumptions about data flow, and tests that pass but do not test what they claim. This is 60% of the job.
• Understands the full-stack surface. React 19 with TypeScript, Vite build system, React Router for client-side routing, Supabase (PostgreSQL + RLS + Auth + Edge Functions + Realtime), TanStack React Query for server state, Zod validation, Tailwind CSS + shadcn/ui component library, Framer Motion for animations, Recharts for data visualization.
• Performance instincts for data-heavy UIs. Virtualization for large datasets (founder content libraries, hook banks, asset lists). Proper memoization (useMemo, useCallback, React.memo). Knowing when to push computation to Supabase Edge Functions vs. handling client-side. Understanding bundle size and code splitting.

AI Workflow

• Daily, production-level experience with AI coding agents. Claude Code, Cursor, Copilot, Codex, Windsurf — any of them. Real usage, not casual autocomplete. You've shipped production features where AI wrote the code, and you ensured it was correct.
• Understands that the AI is the author and the human is the reviewer. Your job is to catch what the AI misses, not to rewrite what it produces. You're the quality gate, not the typist.
• Process discipline. Willingness to always validate through structured review before approving. The validation layer is non-optional. You trust the process more than your gut.

Strong Signals

• Has done a code audit or security review before. Familiar with OWASP Top 10.
• Has been on-call and debugged production incidents under time pressure.
• Has built or worked with Supabase Row Level Security policies or equivalent database-level access control in production.
• Has built or customized AI agent workflows — Claude Code skill files, CLAUDE.md configurations, memory persistence, multi-model comparison, adversarial prompting.
• Has experience reviewing AI-generated PRs specifically (not just AI-assisted coding). Knows what AI gets wrong and why.
• Has worked with TanStack React Query for complex server state management — cache invalidation, optimistic updates, infinite queries.
• Has experience with Playwright E2E testing in a real CI/CD pipeline. Knows the difference between tests that prove the app works and tests that inflate coverage.
• Has built real-time features — Supabase Realtime, WebSockets, or equivalent. Understands subscription lifecycle and reconnection handling.
• Comes from a product SaaS background (not agency, not consultancy). Has owned features end- to-end in a product that serves real users.
• Has experience with content management systems, creator tools, or personal brand platforms. Understands the domain.

Red Flags

• Uses AI only for autocomplete or boilerplate generation.
• Cannot explain the architectural decisions in the code they shipped with AI assistance.
• Treats AI as infallible OR treats AI as untrustworthy. The right stance is: trust but verify, always.
• Overrides correct AI output based on habit or gut feel instead of validating through a structured process.
• Needs to write code by hand to feel productive. The metric here is features shipped, tested, and stable in production — not lines written.
• Frontend-only or backend-only specialist. The review surface is full-stack — React through Supabase Edge Functions through PostgreSQL.
• Has only worked in consulting/body-shop environments with no production ownership.
• No experience with real-time, data-heavy applications. Our platform has complex state, concurrent users, and content generation workflows that require careful architecture.

Who This Person Is

3-7 years of full-stack experience, weighted toward React + TypeScript + Supabase (or equivalent PostgreSQL stack). Not a 15-year veteran — we need someone young, agile, and native to AI-assisted development rather than resistant to it.

Senior full-stack engineers at SaaS companies. People from the React/TypeScript community who also understand databases, auth, and infrastructure. Engineers who have built product features end-to-end —not just components, but complete user flows with backend logic, database schema, auth guards, and deployment.

This person understands the big picture of a production system — how the frontend, API layer (EdgeFunctions), database (PostgreSQL + RLS), auth (Supabase Auth), and third-party integrations (Stripe,ManyChat (social APIs) fit together. They do not need to be an expert in every layer, but they need enough depth across all of them to catch when AI generates something architecturally wrong. And they need enough humility to always validate before approving.

How We Evaluate

We do not use a traditional coding test. We do not need to see you build something in 60 minutes. We need to see you think, review, and direct.

Stage 1: Bug Triage

We give you a real bug report from our codebase (sanitized). Pick one issue and walk us through how you would diagnose and fix it. We are evaluating: Do you read the code? Do you ask the right questions? Do you understand what could go wrong beyond the immediate symptom?

Stage 2: Cold Code Review

We show you a file from our codebase that has not been audited. It was written by an AI agent. Tell us what you see. We are evaluating: Do you catch the security issue? The race condition? The test that passes but does not test what it claims? The Supabase RLS policy that leaks data across tenants? The React Query cache that serves stale data after a mutation?

Stage 3: Task Brief Writing

We give you a feature requirement. Write the task brief you would hand to an AI coding agent. We are evaluating: Is it clear? Is it scoped? Would it actually produce good output on the first pass? Did you think about edge cases, error states, RLS implications, and security boundaries before the agent has to?

Stage 4: AI Workflow Conversation

Describe your current AI coding workflow in detail. What tools do you use? What is your process? What do you trust? What do you double-check? How do you handle uncertainty? We are evaluating: Is this someone who has internalized AI-native development, or someone who is still experimenting?

What We Offer

• A seat at the table. You are joining a founding-level product team, not filling a headcount. Your judgment shapes the product. You work directly with the CEO.
• The most advanced AI development workflow you will work in. Multi-model agent orchestration, AI-generated codebase, structured validation pipelines, Claude Code skills and memory. This is where software development is going — and you will be at the frontier.
• Real production impact from day one. The product is live, founders use it daily, and your work ships to real users immediately. No 6-month onboarding. No waiting for approvals. Ship, validate, iterate.
• A product people love. We are building the operating system for personal brands. 18,000+ founders. $15M/year in revenue. When the product is outstanding, founders stay, tell their friends, and post about it. You are building something that directly changes how people build their businesses.
• A process that values getting it right over getting it fast. We validate everything. We review everything. We do not skip steps. Speed comes from AI; quality comes from you.

To apply or refer a candidate: Send a brief note explaining your current AI coding workflow and a link to something you have shipped. Include your experience with React, TypeScript, and Supabase (or equivalent). No cover letter, no formalities. We read code and we read people — show us both.