Expertise / RAG & Knowledge Engineer
RAG & Knowledge Engineer
„Our knowledge sits in people’s heads and wikis — nobody finds anything."
Builds company knowledge bases that still work in two years. Vector DBs, semantic search, source citation per answer.
Knowledge engineering is more than an embedding endpoint. We design the knowledge architecture so it stays maintainable: source hierarchies, re-index frequency, deletion duties, trust grades per source. The result is a knowledge base that still works in 2 years — not just in the first weeks after go-live.
What clients ask us about this.
- „How do we make our service knowledge usable for AI?"
- „Which vector DB? Local or cloud?"
- „How is source provenance shown in the answer?"
- „What happens when a document has to be deleted?"
What we keep for you.
Documented per mandate, viewable and correctable by you anytime.
- Knowledge architecture with source hierarchies
- Vector-DB configuration and re-index schedule
- Trust grades per source (official / internal note / external)
- Deletion-duty tracking (GDPR + internal retention)
- Hallucination detection patterns from reviewer data
Plant builder · 380 employees · service knowledge base
Knowledge base with 12,000 service tickets + 1,800 technical documents. Source hierarchy: official documentation has highest priority, service tickets provide context. Bi-annual re-indexing, deletion pipeline for departing customers’ data.
One request is enough.
Write us briefly what’s on your plate — we respond within one business day. This role is part of the Autopilot retainer, not booked separately.