Back to Projects
Agentic AIintermediate

HR Policy RAG with GCP

Build and deploy an enterprise-grade HR Policy Assistant that delivers accurate, source-cited answers to employee queries across real company policy documents. The project integrates hybrid semantic and keyword retrieval with reranking, bulletproof AI security guardrails, fallback routing, and serverless deployment on Google Cloud Platform.

22 lectures

What You Will Learn

Architecting a hybrid retrieval pipeline combining dense semantic vectors and sparse BM25 indexing in Qdrant
Implementing advanced reranking using Jina cross-encoders to optimize grounded response generation
Integrating AI security guardrails with Google Model Armor to prevent prompt injections and jailbreaks
Building high-availability LLM routing and automatic fallback architectures using LiteLLM
Evaluating RAG accuracy, faithfulness, and groundedness using automated LLM-as-a-judge pipelines in LangSmith
Containerizing production RAG applications with uv and deploying serverless containers to Google Cloud Run with OAuth authentication

System Architecture

HR Policy RAG with GCP Architecture Diagram

High-level architecture overview of the HR Policy RAG with GCP .

What You'll Build

  • End-to-end multi-format ingestion pipeline converting PDF, DOCX, and PPTX policy files into hybrid vector embeddings
  • Conversational HR Policy Agent equipped with source-citing capabilities, reranking tools, and per-session memory
  • Multi-layered AI defense system featuring Model Armor guardrails, semantic caching, and scope verification
  • Production-ready Google Cloud Run deployment authenticated via Google OAuth with Secret Manager integration

Project Instructor

Divesh Jadhwani

Divesh Jadhwani

3+ years exp
LinkedIn
HR Policy RAG with GCP
Premium
One Subscription. 40+ Projects. Unlimited Access.
AccessMobile & Web