Mohammed Usmani

I’m an AI & ML Engineer based in Bengaluru, India. I write about the practical plumbing required to build agentic AI systems that actually survive in production. Over the last year, I’ve architected and shipped 15+ AI systems, including multi-agent orchestration layers, realtime voice assistants, and hybrid RAG pipelines for sales and accounting platforms. I write primarily about LangGraph, Voice AI (Deepgram/Voxtral), RAG architecture (Qdrant/pgvector), and the FastAPI/TypeScript backends that tie them together. I care more about deterministic tie-outs and predictable workflows than making chatbots sound clever. Currently building at Garage. You can see the full architecture of the systems I've shipped (and talk to the AI agent that knows my codebase) at mohammedusmani.me.