Senior Data Engineer · Compunnel
Every week brings another AI demo that takes the breath away and another enterprise pilot that quietly stalls in production. The gap between the two has very little to do with model quality. It has everything to do with the unglamorous work of data engineering, integration, and governance that surrounds the model.nThis keynote offers a practitioner's view of applied AI, drawn from real enterprise integration work at Fortune-500 scale across financial services and digital media. We will look at why most AI projects stall fragmented data, governance debt, unbounded cost curves and at the modern stack that quietly powers the ones that succeed: iPaaS, event streaming, vector stores, and a thin AI layer sitting on top.nFrom there, we will walk through three patterns engineers can ship inside a quarter: enterprise RAG done right, event-driven AI on a Kafka backbone, and AI-augmented ETL pipelines. Each pattern is grounded in lessons from production chunking strategies, governance gateways, eval discipline, human-in-the-loop design, and the small architectural choices that separate pilots from systems of record.nThe talk closes with a practical learning roadmap for engineers preparing to move from SQL and data platforms into GenAI, and a central thesis worth carrying home: in the age of applied AI, the data engineer is the new AI engineer. Models will keep getting better. The work of moving the right data to them safely, on time, with a clear contract is what unlocks intelligence at scale.