# Senior React + FastAPI Developer — Take AI-Powered SaaS from Demo to Production ## The Problem I have a working prototype of an AI-powered SaaS product built with React 18 + TypeScript + Tailwind v4 on the frontend, FastAPI on the backend, and Supabase for auth and database. It uses the Anthropic Claude API to generate dynamic UI components from data. The product works at demo level but is not production-ready. I need an experienced developer to fix the engineering problems below and get it to the point where real users can rely on it daily. ## Tech Stack - React 18 + Vite + TypeScript + Tailwind v4 - FastAPI (Python) - Supabase (PostgreSQL + Auth + Storage) - Anthropic Claude API - Recharts (data visualisation) - Currently on localhost — needs production deployment ## Technical Challenges ### 1. Unstable Data Pipeline - The app sends user data context to the Claude API for AI generation - Data context intermittently fails to reach the API call — suspected race condition between auth state, data fetching, and the chat send path - Chat state loses context across turns — the AI sometimes responds without awareness of the uploaded data - Need this diagnosed and fixed so the pipeline works reliably every time ### 2. AI Output Quality - Claude API generates JSX that renders inside an iframe via srcdoc - The quality of the generated output is significantly lower than what the same model produces in direct conversation - Issues include: inconsistent styling, broken chart labels, incorrect number formatting, poor colour control - This requires a combination of prompt engineering (improving the system prompt and context sent to the API) and frontend component work (Recharts configuration, consistent CSS) - Must enforce strict design rules: specific colour palette, consistent card styling, correct number formatting (currency, percentages, counts), proper chart axis alignment ### 3. Google Sheets Integration - Build a connector that lets users authenticate with Google and select a Sheet as a live data source - Data should refresh on a practical schedule (polling is fine) - Google OAuth flow, Sheets API v4, token storage in Supabase - Must handle common edge cases: empty columns, mixed types, large sheets ### 4. Production Deployment - Frontend to Vercel or Netlify - Backend to Railway or Fly.io - Supabase is already cloud-hosted - Environment variables, CORS configuration, SSL, domain setup - Must be stable — not just deployed but monitored and reliable ## What's NOT in Scope - No new features — this is about making what exists work properly - No redesign — the UI is built, it just needs the engineering underneath to be solid - No additional data connectors beyond Google Sheets (for now) ## What I'm Looking For ### Must Have - Shipped at least one React + Python/FastAPI product to production - Experience with Supabase or similar (Firebase, Auth0 + Postgres) - Experience calling LLM APIs (OpenAI or Anthropic) from a backend and rendering the output in a frontend - Experience with Google OAuth and Google Sheets API - Strong debugging skills — most of this work is diagnosing and fixing, not building new ### Nice to Have - Experience with Recharts or D3 - Experience with Tailwind v4 - Familiarity with Anthropic Claude API specifically ### Working Style - Close collaboration — daily or every-other-day async updates - I will provide detailed specs, test cases, and context on every issue - I want to understand what you're doing — not just hand it off