scaling.md

Scaling Guide

• Switch database_url to PostgreSQL for multi-worker deployments.
• Increase queue.max_parallel_jobs for higher throughput.
• Move artifact storage to a high-capacity volume or object-store-backed plugin.
• Add provider configurations for local GPU-serving endpoints such as vLLM or Ollama.
• Replace the default vector store with a plugin-backed FAISS, Chroma, or Qdrant adapter as the dataset grows.