Enterprise Knowledge Retrieval Assistant (RAG)
- RoleAI Architect
- ClientPublic sector, Australia
- PeriodOct 2024 - Feb 2025
<1.8s latency RAG assistant; ~30% reduction in Tier-2 escalations.
Data
Policy manuals, SOPs, FAQs, and internal documents across PDF, DOCX, and HTML.
Objective
Deploy a secure retrieval-based assistant to answer internal queries with traceable, source-linked responses.
Outcomes
- Delivered a production RAG system with source-linked responses and feedback loops.
- Achieved <1.8s end-to-end latency with hybrid retrieval and prompt-side caching.
- Reduced Tier-2 escalations by ~30% through accurate, attributable answers.
Tech
Azure OpenAI, Azure AI Search, Python, LangChain, Azure Functions, Blob Storage.
Stack
- Azure OpenAI
- Azure AI Search
- Python
- LangChain
- Azure Functions
- Blob Storage