Architecture, React and enterprise practice on Microsoft Power Pages.
Technical deep dives, security boundaries and proven implementation patterns from real client projects.
I could have built Ollama directly into the backend. I could have built OpenAI directly into it. Instead, I built an abstraction layer that allows both, and understood why this is not overengineering but the only strategically sound decision.
Retrieval-Augmented Generation is not theory. It is a concrete three-step architecture: embed, retrieve and generate. I show how I implemented it entirely locally with Qdrant, nomic-embed-text and llama3.2, without cloud costs.
A Chrome Extension for Instagram metadata was the starting point. It grew into a complete RAG system built on Qdrant, Ollama and TypeScript that runs locally, is cloud-ready and enables semantic search across captured social media data.
Why traditional Power Pages frontends with a global jQuery context reach their limits over time, and how a hybrid React architecture enables reliability.
Why choose between low-code and React? Learn how to embed React components in Power Pages and deliver a modern user experience.
Learn how to replace inefficient manual uploads in Power Pages with an automated deployment script and a side-by-side architecture.
Describe your workflow. Together, we will identify the right next step.
Discuss your workflow