New Build End-to-End RetireesParadise

I am spinning up a green-field SaaS that serves many facilities under one roof, so every line of data must respect facility-tagged, row-level security in PostgreSQL via Supabase. The stack is already chosen; what I need is someone who can put all the pieces together and ship production-ready code.

Frontend
– Next.js (React + TypeScript) with user authentication, real-time updates, and a rich analytics dashboard. I expect optimistic UI, WebSocket/Subscribers (Supabase or similar) for live data, and a crisp component library (MUI, Tailwind or your suggestion).

Backend
– A Node.js API (also TypeScript) that exposes REST endpoints to the frontend, enforces multi-tenant rules, and triggers n8n webhooks for SMS, email, and in-app notifications.
– A separate Python FastAPI microservice dedicated to AI work: text generation and broader natural-language processing powered by OpenAI and Claude, implemented with Retrieval-Augmented Generation so we can feed domain documents for higher accuracy.

Data
– PostgreSQL on Supabase with schema migrations, RLS policies, and automated seed scripts.

Hosting & DevOps
– AWS deployment (ECS or Lambda for Node, ECR for FastAPI, S3/CloudFront for static assets), wired into GitHub Actions for CI/CD. I should be able to push to main and watch the entire suite test, build, and ship.

Automation
– n8n flows you configure for outbound SMS, email, and internal notifications using the backend’s webhooks.

Deliverables
• Responsive Next.js app with auth, live updates, and analytics dashboard
• Node.js API with full CRUD, RLS enforcement, Swagger docs, and unit tests
• FastAPI AI service with RAG pipeline, OpenAI/Claude integration, and test suite
• n8n workflow JSON with clear triggers and actions
• Infrastructure as Code (Terraform or CDK) and GitHub Actions pipelines
• README plus a short Loom video walkthrough explaining setup and deployment

Acceptance Criteria
1. A fresh AWS account can be cloned, configured, and deployed end-to-end with one command.
2. Switching facility IDs in auth token immediately scopes all queries and real-time feeds.
3. AI endpoints return contextual answers using supplied reference docs with ≥90 % relevance in provided test set.
4. All critical paths covered by automated tests and passing in CI.

If this sounds like your wheelhouse, tell me how you’d tackle the multi-tenancy layer first, then outline your preferred AWS deployment pattern.