Private GPT LLMs: Securely Deploying Azure OpenAI Service with Terraform
Event:BASTA! Spring 2024
Type:Talk
Date:February 12, 2024
Duration:1h
Location:Frankfurt, Germany
Language:German
Securely deploying Azure OpenAI Service with Terraform for private GPT LLMs and stream results to Angular frontends.
Event Details
Microsoft’s Azure OpenAI Service provides the capability to integrate pretrained Large Language Models (LLMs), such as GPT-4 Turbo, into applications and business processes. However, beyond programmatic access, important questions arise around deployment and hosting:
- How can the Azure OpenAI Service and the desired LLMs be provisioned declaratively using Terraform?
- How can the service be integrated into a virtual, private network infrastructure?
- How can access to the LLM be controlled and restricted?
- What about data residency and data sovereignty — can it be guaranteed that data remains within the EU or Germany?
- And how can response streaming be implemented for frontend applications?
This talk addresses all of these questions, enabling you to determine whether the Azure OpenAI Service is a suitable option for your organization and how it can be effectively integrated into existing or new cloud infrastructures.