Rentoso

Where I first put LLMs into production. I built tool-calling UIs — the layer where a model's actions become visible, interruptible and trustworthy to a human — and the retrieval infrastructure behind them: embeddings, a vector database, and a RAG pipeline that grounded answers in the product's own data.

Equal parts UX judgment and plumbing, which is exactly what applied AI turns out to be.

Stack: Vue/React, Node, embeddings, RAG, prompt engineering.