← All projects
WeCloudData capstone2025

RAG-powered AI chatbot on Azure

The capstone project of my WeCloudData Cloud Engineering bootcamp.

2 deployment architectures · 100% Infrastructure as Code · CI/CD
2deployment architectures
100%Infrastructure as Code
01 The problem

Answering questions accurately from a set of PDF documents, and running it on cloud infrastructure that is reproducible and secure.

02 What I built
  • A retrieval-augmented generation backend in FastAPI with a Streamlit front end.
  • Two deployment architectures, VM Scale Sets and Azure Container Apps, both provisioned with Terraform.
  • Azure Blob Storage, Cosmos DB, PostgreSQL and Key Vault for data and secrets, with CI/CD through GitHub Actions.
03 Results
  • The full stack can be rebuilt from code, with secrets kept out of the codebase.
04 Tech stack
AzureTerraformFastAPIStreamlitCosmos DBKey Vault
# rag-chatbot/main.tf
resource "azurerm_container_app"
    name = "rag-api"
    replicas = 2
Selected projects