LennyVerse

AI-powered product wisdom platform built as a monorepo.

Monorepo Layout

• backend/ - FastAPI service scaffold
• frontend/ - Vite + Tailwind app scaffold
• data-pipeline/ - ingestion, schema migrations, and verification scripts
• data/lennys-newsletterpodcastdata/ - canonical corpus directory

Phase 1 Quickstart

1. Copy env:
   • cp .env.example .env
2. Install Python deps:
   • make setup
3. Normalize dataset location:
   • make normalize-data
4. Apply database schema:
   • make migrate
5. Run ingestion:
   • make ingest
6. Verify ingest:
   • make verify

Common Commands

• API server: make run-api
• Frontend dev server: cd frontend && npm run dev
• Dry-run ingest (no writes): make ingest-dry-run
• Limited ingest dry-run (first N docs): make ingest-dry-run-limit LIMIT=10
• Limited ingest (first N docs): make ingest-limit LIMIT=10
• Tests: make test

Production Deployment

• Production uses a single container (ghcr.io/<owner>/lennyverse) that serves:
  • FastAPI API endpoints under /api/*
  • Built frontend assets from frontend/dist
• This same-origin setup removes cross-origin calls in production and avoids CORS deployment issues.

Operational Runbook (Phase 1)

• Malformed frontmatter: parser logs warnings and skips bad files; re-run after fixing source files.
• Supabase paused / cold start: retry make migrate or make ingest after project wake-up in dashboard.
• Embedding endpoint timeout: ensure Ollama/NIM is reachable at OLLAMA_EMBED_BASE_URL, then retry ingest.