Database & RAG

PostgreSQL + pgvector layer of the FCA support agent. Each sub-page is verbatim from full_project_context_updated.txt:

• SQLAlchemy Async Pooling — environment-aware NullPool vs pool_size engine
• Alembic Migrations (Sync/Async Bridge) — sync alembic env.py calling async run_async_migrations() with AsyncSessionLocal
• pgvector RAG Ingestion — CREATE EXTENSION vector then load-and-embed PDF chunks
• RAGService ingest_pdf End-to-End — PDF → chunks → embeddings → pgvector upsert, with tqdm progress