Most RAG chatbots don’t fail because of bad models — they fail because the system was built in the wrong order. This post explains the correct build sequence that actually survives production
December 30, 2025
Most Retrieval-Augmented Generation (RAG) systems look impressive in demos — and then quietly fail in production. This post explains why, and what actually breaks after the first real users arrive.
Creating content just for reach is a trap. I’m turning my engineering output into products — and using content to stress-test them in public.
July 25, 2025
FastAPI (or Django) + Next.js, Sanity for CMS, Groq/Llama-3 or OpenAI for LLMs, and a retrieval layer that I can benchmark, observe, and replace.
Why I moved my blog under my name, what I’ll write about (AI engineering, RAG, agents, shipping products, and content), and how I’ll keep it honest and useful.
