Skip to content
Emmanuel NanadoumRésumé(PDF, opens in a new tab)
Enterprise AI implementation
KnowledgeOS

Grounded answers. Visible evidence.

Ask the company.
See the source.

A portfolio implementation of retrieval-augmented generation. Questions are retrieved against a labeled sample knowledge base, then the answer is constrained to the retrieved evidence and returned with citations.

Knowledge base
Handbook · Security · Sales · Implementation
Retrieval
Gemini embeddings with lexical fallback
Generation
Grounded Gemini response
Guardrail
Unsupported answers are refused
Knowledge queryready
Grounded response

Ask a question to run retrieval, grounding and citation generation.

01

Retrieve

The question is embedded and compared with the knowledge chunks.

02

Rank

The most relevant evidence is selected before the model sees the question.

03

Generate

The model is instructed to answer only from retrieved evidence.

04

Cite / refuse

Sources are shown. Unsupported questions receive an explicit refusal.