Hey HN,

I built Rivestack because managed PostgreSQL with pgvector is either expensive (AWS, GCP) or shared resources (Neon, Supabase free tiers).

What it does: - Dedicated PostgreSQL with pgvector pre-installed - Automated backups, monitoring, HA - EU and US-East regions - Free tier for testing

Benchmarks on the $29 tier: - 10k vectors: 2,000 QPS, <4ms latency - 1M vectors: 252 QPS, 32ms latency, 98% recall

Stack: Hetzner infrastructure, Patroni for HA, pgBackRest for backups.

Would love feedback from anyone building RAG or semantic search. Happy to answer questions!