What 30 years of working in Japanese taught me about engineering leadership
By the time you have worked in Japanese for thirty years, fluency stops being about vocabulary and becomes an operational capability for leadership.
Read essay →What building solo systems after hours taught me about leading engineering teams
When you build alone, there is nobody to misunderstand your requirement and nobody to rescue your architecture. The feedback loop is merciless and useful.
Read essay →Why I run my own AI agent infrastructure — and what it teaches me about enterprise AI
My AI agents save me time some days and create debugging work on others. That tradeoff is exactly why they are worth running.
Read essay →The difference between an AI demo and an AI product
I have seen many AI demos that looked inevitable and very few AI products that deserved to exist in production. The gap is where most investment disappears.
Read essay →From biosensors to LLMs: 20 years of measuring things that are hard to measure
In 2003 I was measuring lipid bilayers on silicon microelectrodes. Two decades later I was measuring failure behavior in AI systems. The tools changed. The measurement discipline did not.
Read essay →What ISO 13485 actually means for software engineers
When engineers say they already do most of ISO 13485 informally, they are often right. The problem is that informal discipline cannot be audited or scaled.
Read essay →17 POCs before we shipped: how we validate AI systems in regulated healthcare
Validation in healthcare AI is expensive, but failed production deployments are usually more expensive. That is why deliberate POC discipline matters.
Read essay →How we built a $16M digital-health spin-off inside a 90-year-old Japanese company
The hardest part of building HeartVoice was not the technology. It was persuading a hardware-first organization to fund a data-platform future without destabilizing its current business.
Read essay →Six countries, six data-privacy laws: what engineering leaders need to know about ASEAN health data
There is no single ASEAN privacy problem. There are several jurisdiction-specific problems that share some vocabulary and collide inside the same platform.
Read essay →Why the POC phase is the most misunderstood tool in corporate innovation
The seventeenth POC was the one we shipped. The other sixteen were not waste. They were the cost of learning what a production system would actually have to survive.
Read essay →How to ship AI in a regulated healthcare environment without killing your velocity
Shipping AI in healthcare is difficult for reasons that have very little to do with model demos and a great deal to do with traceability, risk, and operational discipline.
Read essay →Distributed engineering across five time zones: what actually works
The time zones are manageable. The dangerous part is assuming that people in different engineering cultures interpret the same instruction the same way.
Read essay →Hypothesis-driven engineering: applying the scientific method to product decisions
Every product decision begins as a claim about the future. Teams that forget that build faster, but they do not necessarily learn faster.
Read essay →Why Japan's monozukuri philosophy makes you a better engineering leader everywhere
Monozukuri is often translated as making things. In practice it is closer to building with respect for process, mastery, and the long-term cost of shortcuts.
Read essay →What a PhD in physics taught me about leading engineering teams
The most useful thing my physics Ph.D. taught me about leadership was not a technical technique. It was a disciplined way of moving from uncertainty to evidence.
Read essay →