Skip to main content
MDSDIG
Back to home

SYMBAN — Main case

Translation in progress. The full case study (architecture, stack, lessons learned) is available in German at michailidou-services.de/cases/symban.

In a nutshell

SYMBAN is my own AI product: a multi-pass pipeline + multi-index RAG (with pgvector) for novel production. From research to draft chapter in a deterministic flow — with quality gates after every pass.

What you'll find in the German version

  • Architecture: multi-pass pipeline + multi-index RAG (pgvector)
  • Stack: Python, PostgreSQL/pgvector, Next.js, OpenAI / Anthropic
  • Stage layout: 9 passes, each with its own quality gate
  • Lessons learned on retrieval, drift control, and pass orchestration

For the full version, please consult the German case study.