Writing
Notes on event streaming, enterprise integration architecture, and building reliable systems — drawn from 13+ years in the field. No fluff, no listicles.
Most Kafka migrations fail for organizational reasons, not technical ones. Notes from leading an events-streaming practice across large enterprise teams — on schemas as contracts, the pull toward point-to-point topics, and why 'real-time' is usually the wrong goal.
A few years ago I helped build and patent a system that scans code for known vulnerabilities and automatically fixes them. The AI/ML part was the least interesting piece — the knowledge base and the feedback loop were what actually made it work.
From consolidating two companies' inventory systems after a merger to migrating a decade of webMethods interfaces onto Oracle Fusion Middleware — the failure mode is almost never the target platform. It's the integration nobody knew existed.