Writing

Essays on model-based systems engineering, traceability, verification, and what it takes to let automated tooling near a safety-critical model. Mostly things I wish someone had written down before I learned them the slow way.

• Why Requirements Traceability Breaks, and What Actually Fixes It: Traceability decays on real programs for structural reasons, not because engineers are lazy. A look at the actual failure modes and what a fix has to do.
• Model Governance for SysML Programs That Have to Certify: When the model is the certification artifact rather than a picture of one, governance stops being bureaucracy and becomes the thing that makes the model trustworthy.
• Interface Control Documents Should Be Models: The ICD is where integration problems are born. Treating it as a document instead of a model is why interfaces that pass review still fail on the bench.
• Guardrails for AI-Generated Engineering Artifacts: Reviewing generated requirements and models by reading them does not scale and does not work. The check has to be structural, and the generator must not be able to route around it.
• Model Context Protocol for Engineering Toolchains: Why exposing MBSE context through typed, governed operations beats dumping exported documents into a context window, and what the interface design has to get right.
• Deriving Safety Assessments From the Model Instead of Maintaining Them Beside It: Hazard analyses go stale because they are maintained in parallel with the architecture. Generating them from the model changes which errors are possible.
• What I Learned Leading an MBSE Community of Practice: Running a 100-engineer MBSE community of practice at Boeing taught me that modeling adoption fails for organizational reasons, and that the tool is almost never the problem.
• DoDAF and SysML on Defense Programs, Without the Viewpoint Theater: DoDAF viewpoints get produced because they are deliverable, not because anyone reads them. Making them fall out of a coherent model is what turns compliance into engineering.
• Modeling Automotive HMI and Infotainment Integration: Notification timing, media routing, and vehicle state synchronization defects are integration problems disguised as feature bugs. Behavioral modeling is how you find out which is which.
• From the Factory Floor to Systems Architecture: How writing overhaul instructions for flight control actuators turned out to be useful preparation for architecture modeling, and what I would tell someone making the same move.