Backend / AI Engineer: Supabase Edge Functions, Postgres, & LLM Structured Extraction

MAKE SURE TO READ THE ATTACHED FILE We are building mobile app for US compliance and renewal tracking platform for licensed tradespeople (list below). The foundational core of our product is highly accurate, multi-dimensional regulatory rule data (comprising cycle types, anchors, subject sub-buckets, carryover limits, and reporting models). Crucial Architecture Requirement: We never write AI-generated data directly to our production rules database. Our LLMs make mistakes; state boards do not. We are hiring an expert backend developer to build an Evidence-Collector Agent that automatically watches state board web pages, detects text changes, runs the diff through an LLM to extract structured data, and queues a proposed payload with verbatim source quotes to a restricted admin audit queue for human approval. Our Tech Stack Frontend: React / TypeScript / Tailwind (built via Lovable.dev) Backend & DB: Supabase (PostgreSQL, Auth, Row-Level Security, pg_cron) Runtime: Supabase Edge Functions (Deno / TypeScript) Scraping Layer: Firecrawl API (or similar markdown-conversion API) LLM Engine: OpenAI API (GPT-4o/GPT-4o-mini using JSON Schema / Structured Outputs) or Anthropic Claude API