I build production LLM systems — not demos. At Iris Software I engineer a conversational AI platform for banking: intent classification, slot filling, agentic API execution, and RAG-grounded responses averaging under 5 seconds. Under the hood: 4-tier model routing on AWS Bedrock (Nova → Claude Sonnet), RAG on PostgreSQL + pgvector with hallucination guards, and a provider-agnostic LLM layer (Bedrock ↔ OpenAI in one config change). Before GenAI, I spent 3+ years building ETL pipelines and cloud data platforms with PySpark, Databricks, Azure, and AWS — so my AI systems ship with real data foundations. Currently pursuing an MS in AI & ML (Woolf). 13k+ engineers follow my AI content on LinkedIn. 🎯 What I can help you with: • Career transition into GenAI/AI engineering — the exact path I took from data engineering • AI engineer interview prep — LLMs, RAG, vector DBs, agents, system design • Your LLM project — architecture review, RAG debugging, cost optimization, agent design • Teams/founders — scoping and building production GenAI features on AWS 🛠 Stack: Python · FastAPI · LangChain · AWS Bedrock (Claude, Nova) · pgvector · ChromaDB · PostgreSQL · MCP · Databricks · PySpark Why me: I made the transition you're trying to make — support → data engineering → GenAI — and I now ship the systems interviewers ask about. Book a call if you're transitioning into GenAI, prepping for AI engineer interviews, or building LLM products.