Semantic Caching Demo + Cost & Sizing Calculator (Azure Managed Redis)

This project is a full working application that demonstrates practical semantic caching patterns for AI workloads: a user-visible, real-time streaming experience paired with a cost model that helps decision-makers understand the tradeoffs between repeated LLM calls and Redis-backed reuse. The solution uses a Next.js frontend and a FastAPI backend, with Redis Stack as the semantic cache store. It also includes an Azure deployment path that provisions Azure OpenAI, Azure Managed Redis, and Azure Container Apps using azd.

• Building a demo that feels like a real app (streaming UX), not a toy example.

• Turning “semantic caching is good” into a measurable business case via cost & sizing modeling.

• Packaging infra + app deployment so others can reproduce it quickly in Azure.

• Shipped a deployable reference app that showcases semantic caching and streaming behavior end-to-end.

• Added an interactive calculator to support sizing conversations and justify spend using workload economics.