Systems engineering · AI · machines in the real world
I work across distributed software, AI-assisted engineering, robotics, and intelligent machines. The common challenge is turning powerful technology into systems we can understand, verify, operate, and trust. Twenty-plus years of production engineering are the foundation; the physical world is the newest proving ground.
Engineering perspective
My work has crossed web platforms, advertising technology, large-scale data systems, forecasting, cloud infrastructure, and applied AI. Robotics does not replace that history. It stress-tests the same architecture and reliability instincts in a world where sensors, batteries, motors, networks, and gravity all get a vote.
Connected domains
The technology changes. The hard questions—boundaries, failure modes, verification, ownership, and operations—remain surprisingly consistent.
More than two decades across APIs, data pipelines, forecasting, cloud platforms, performance, reliability, and operational tradeoffs—including current work on forecasting and monetization systems at Microsoft.
See current work on LinkedIn →AI can generate more software than a team can casually review. The durable advantage becomes judgment: deciding what should exist, how to prove it works, and who owns the result.
Hands-on work with ROS 2, robot networking, telemetry, power systems, perception, and recovery—treating the robot as a complete system rather than a collection of demos.
Follow the field notes →W3C Immersive Web API review, patented client-side content techniques, ad tech, analytics, video, and Apple TV work—different surfaces connected by the same interest in systems that remain understandable as they scale.
Systems, AI & Machines journal
Decisions, experiments, failures, measurements, and changing assumptions across software systems, AI, robotics, and machines operating in the real world.
A programming book from 2003 predicted the strange part of the AI era: coding gets easier while systems get harder. What becomes valuable now is judgment, verification, and ownership.
AI can write 14,000 lines before lunch. Someone still has to decide whether those lines should exist.
Working principles
Connect
LinkedIn carries the professional timeline. The journal holds the reasoning and evidence. GitHub and X carry artifacts, shorter updates, and technical conversation.