RAG AI assistant for ISO quality documentation
Local PoC to query a mock ISO documentation corpus with cited answers, confidence scoring, document alerts, and inconsistency detection.
Context
In quality and audit contexts, quickly finding the right information, checking document applicability, and spotting inconsistencies can be time-consuming and sensitive.
Problem
Manually reviewing a document corpus leads to delays, interpretation risks, limited answer traceability, and difficulty spotting documents under revision.
Solution
I designed a local RAG AI assistant that retrieves relevant passages from a document corpus, generates a controlled response, displays sources, and flags cases requiring human validation.
My role
End-to-end PoC design and implementation: corpus structuring, RAG pipeline, confidence logic, Streamlit interface, interaction logging, and business framing focused on auditability.
Visuals
Stack
Highlights
- Cited answers with document excerpts and cautious out-of-corpus refusal behavior.
- Explainable confidence score with alerts for documents under revision and version mismatches.
- Interaction history and logging to improve business auditability.
Discuss a similar need
I can contribute to Power Platform, SharePoint, Power BI, and Microsoft 365 automation initiatives with a field-oriented, execution-focused, and adoption-driven approach.
Send an email Back to projects