I fix processes and write the software that keeps them honest; I prefer calm systems to heroic saves.
I did not start with code and I actually prefer not to; I started with shifting schedules, timers that nagged, rework that circled back, and deadlines that leaned forward. I learned that if people sprint every day, the process is broken; if the same fire returns every week, it is not a fire, it is policy. So I studied flow, watched where time pooled, and quit firefighting. Software came later; it became the tool that let a better process survive contact with Monday morning.
Origin; why process first
My path was from the lab floor to code. Move product; sample when needed; remake what fails; repeat until the pattern becomes obvious. Competence tempts you to coast, heuristics carry you, and then the volume doubles while headcount does not; the old tricks buckle.
That is when I reached for the Toyota Production System. Put the work on the wall; name the waste; restore flow. Do the simplest change that moves the bottleneck; make it easy to keep doing. Sometimes the fix is a new route through the room; sometimes it is a counter, not a keyboard. The easiest task is the one you no longer do.
Value stream maps gave me a clear view; constraints taught me discipline; small problems taught me to leave no sharp edges. I try to make changes that do not need a meeting to stay changed. I also learned the hard lesson that not every good idea survives reality; experiments fail fast, notes matter, and the team keeps the change alive.
What I build; what I refuse to build
I build internal platforms that reduce variance and make work predictable. Lightweight services with clear contracts; data models that express the questions people ask; floor‑facing screens that do not play cute. Pipelines that refresh on predictable beats; audit trails a skeptical auditor can follow. I refuse to build magic boxes. If the tool hides the process, the process rots. I build with teams, not alone; the best tools come from operators, engineers, and managers iterating together.
How I work; the short rules
Ship small; measure; adjust. Keep one source of truth; write the decision next to the code; delete stale ceremonies. Prefer clarity over cleverness; prefer logs over legends; prefer repair over blame. I work async and write crisp updates; I live in Japan and can line up with Europe or the United States without wrecking sleep schedules. I welcome constraints; they keep the design honest. I am wrong sometimes; metrics correct me and tests keep the mistakes small.
Manufacturing creed; Lean without the posters
I keep the kanban current and write standard work that anyone on the team can use. Kamishibai checks save more than they cost when they are simple to run and impossible to fake. Flow is the point; inventory is a story we tell ourselves when we do not trust the next step. If an operator needs a spreadsheet to remember the job, I failed the design. If a lot is late or fails QC, the process failed; when I own the process, that is on me. If an associate cannot finish on time or a step is unclear, I fix the system before I judge the person. I strive to hand work back to the team in a form they can sustain.
Lessons learned
- Measure before you change; assumptions are expensive.
- Ship an experiment, not a sermon; small failures cost less and teach faster.
- Documentation is kindness to the future person who inherits the work.
- Involve the people who do the work early; their constraints will be the constraints you must live with.
- Ownership beats blame; name the owner and the signal that work is done.
Science bones; why I am fussy about data
Biochemistry and DNA work made me meticulous; instruments do not care about intention. A lazy calibration becomes a failed run; a missing unit becomes a ghost in the database six months later. That experience taught me to protect data at the edges; validate early; label everything; never let a measurement travel without context. When in doubt, measure twice; when certain, measure anyway. Always write the units; always label the axes.
Architecture; simple on purpose
I prefer fast APIs that tell the truth; tables with names that mean what they say; indexes that earn their keep. PostgreSQL for the spine; materialized views where the questions recur; background jobs that run on a schedule. Front ends that fit the hands that use them; buttons that tell you what will happen before you click. Observability is not an afterthought; it is the way a system speaks when something hurts. I favour pragmatic choices that teams can support without drama.
A few field notes; small stories, cleaner lessons
- We cut a queue with a two‑hour change: one column, one rule, one screen; the backlog slid downhill for a week and never returned.
- A team swore a task took five minutes; the log said nineteen on odd days and forty on deadline days; we changed the handoff, not the people.
- A nightly job failed twice a month; the fix was not a retry, it was a contract: do not publish partial data. The alert disappeared and the pager grew silent.
Principles; written where I can see them
- Own the process; own the outcome; leave the sharp edges round.
- Prefer nouns to verbs; prefer facts to feelings; prefer deletion to decoration.
- Tests and logs before cleverness; comments that say why, not what.
- One source of truth; one owner; one path to change.
- Boring is a compliment; uptime is a love language.
Personal; the human who builds the machines
I live in Japan with my family; I collect goshuin and bike + hike to temples that lean into the mountain; I keep a compost bin that reminds me everything becomes soil if you are patient. I write small games for fun; Dragon Quest brings me pure joy. Nature insists on feedback loops; software should do the same. I am a parent and a partner; those roles keep me honest and keep priorities clear.
Working with me; expectations in plain words
Expect small, frequent changes; clear notes; and problems called by their names. I will push for a single source of truth; I will ask for ownership that fits in a sentence; I will trade clever for clear. If we disagree, we will test; if we agree, we will still measure. I will admit when I do not know; I will ask for help and I give credit when the team fixed what I could not.