Skip to content

Latest commit

 

History

History
70 lines (40 loc) · 2.96 KB

README.md

File metadata and controls

70 lines (40 loc) · 2.96 KB
Raw

Google Cloud Infrastructure for Document Question Answering with Generative AI and Elasticsearch

This repository contains the Google Cloud infrastructure deployment for the llmdoc project, which implements a document question answering system using Generative AI and Elasticsearch.

Pipeline Design Principles

Our development and deployment pipeline adheres to the following principles:

  1. Rapid Iteration: All dependencies are kept within the project to maximize short feedback development cycles.
  2. Seamless Deployment: End-to-end deployment and testing can be executed with a single command: make all.
  3. Configuration Management: Deployment target differences are managed through google_project.tfvars configuration files.
  4. Version Control: Code base changes are tracked using git branch, while deployment states are tracked with git tag.

Deployment Architecture

Deployment Diagram

Our deployment stack leverages various Google Cloud services and open-source tools to create a robust and scalable infrastructure.

Security Static Analysis

We use Checkov, a static code analysis tool, to scan our infrastructure as code (IaC) files for potential security misconfigurations or compliance issues.

To run the security analysis:

make checkov

Checkov Analysis Demo

Infrastructure as Code with Terraform

We use Terraform to manage and provision our Google Cloud infrastructure. This allows for version-controlled, repeatable deployments across different environments.

To apply the Terraform configuration:

make terraform

Terraform Deployment Demo

Kubernetes and HELM

HashiCorp Vault

We use HashiCorp Vault for secrets management, providing a secure and centralized solution for storing and accessing sensitive information.

To deploy Vault:

make vault

Vault Deployment Demo

Vault Secrets Operator

The Vault Secrets Operator is a Kubernetes operator that synchronizes secrets between Vault and Kubernetes. This allows for seamless integration of Vault's secret management capabilities with Kubernetes applications.

For more information, see the Vault Secrets Operator documentation.

Document Question Answering

The document question answering system deployment is currently in progress. This component will leverage Generative AI and Elasticsearch to provide intelligent responses to queries about document content.

Functional Testing

Comprehensive functional testing of the deployed infrastructure and applications is currently under development. This will ensure the reliability and correctness of the entire system.