No dedicated repository, and the AI Infra Control Plane does not implement RAG. This page documents the retrieval / citation pattern behind the site assistant; the policy, identity and audit layer a governed AI platform needs is proved in the AI Infra Control Plane.
Case 08 · Pattern
RAG Knowledge Platform
A RAG knowledge platform turns repositories, runbooks, architecture notes, and project metadata into retrievable engineering context with citations and safer answer boundaries.
00
Evidence home
Pattern case — proof lives next door.
01
How it works
From curated corpus to a cited answer.
RAG Knowledge Platform by Andrey Lesnikov — export, index, retrieve, and cite engineering context for the AI Twin and operator questions.
-
01
Curate the export
Repos, cases, runbooks, and recognition data become structured markdown — quality over dumping every chat note.
-
02
Index for retrieval
Chunking, embeddings, and freshness rules keep the corpus small enough to trust and large enough to answer.
-
03
Retrieve with bounds
Top-k retrieval stays inside public access boundaries so answers do not invent private context.
-
04
Cite and answer
Responses carry source links so conference and site assistants can be checked, not just believed.
02
Evidence
From problem to proof.
How the system is framed, operated, and proven — not a slide deck summary.
01
Engineering knowledge usually lives across GitHub repositories, runbooks, tickets, docs, and chat history. A generic assistant cannot answer infrastructure questions reliably without curated context.
02
The platform exports structured markdown, project metadata, case studies, and recognition data into retrieval-ready documents with citations and curated local fallbacks.
03
RAG quality depends on curation, chunking, freshness, and access boundaries. A smaller high-quality knowledge base is more useful than dumping every note into embeddings.
Constraints
Source freshness, citation quality, chunking, access boundaries, hallucination control, and explainable answers.
03
Related
Nearby systems.
AI Infra Control Plane
Private AI platforms need shared governance across identity, policy, audit, cost, and SLO signals — not scattered service configs
Private AI with explicit platform contracts
LLM Infrastructure Runtime
LLM workloads move faster than traditional platform controls and can quickly become expensive, opaque, and hard to operate
LLM serving with operability signals
Cloud-Native AI Gateway
AI usage needs routing, policy, budget awareness, and provider resilience
One boundary: rate, budget, fallback, OTel
Available for meaningful infrastructure conversations