Meet Mark

I build AI systems for a living. My doctorate is in economic geography, the study of why some places prosper while others don't. I haven't worked as a geographer for years. The question stayed anyway. AI is the technology most likely to change the answers in our lifetime, and this site is where I think about that combination in public.

I was born in Japan and moved to the United States in my twenties. In thirty-some years here I have lived in New York, Oklahoma, California, and Hawaii, and now California again. The moving around wasn't a plan. It's where the work was.

What I do now

I have a day job in AI, and this site is not about it. The time that is mine goes to KahiDreamers, a one-person company for my civic projects. The largest is Datatrove (datatrove.ai). It indexes 71 Hawaii government and research sources (more than 11,000 public datasets) and answers plain-language questions about them, with the sources attached. A California version is underway, at 60 sources and more than 2,800 datasets as of early July 2026.

This spring I lectured on AI and ethics in a public-policy graduate course at the University of San Francisco.

How I got here

I trained as a mechanical engineer at Keio University (my master's thesis measured how fine fuel drops evaporate), then switched fields for a Ph.D. in Regional Science at Cornell, with a summer fellowship at the Santa Fe Institute's Complex Systems Summer School along the way. My dissertation was about agent-based models, which simulate how many small decisions add up to large patterns. Most of my career since has been some version of that question. While at Cornell I also helped a Harvard Medical School group simulate tumor growth with the same tools (cells cluster; so do people).

After Cornell I spent a decade in industry. I optimized store locations in Tulsa, automated hundreds of daily sales forecasts for a convenience-store chain, and built TV advertising optimizers and brand-linkage methods in Los Angeles. In 2011 I moved to Hilo as a postdoctoral researcher at the University of Hawaii. There I adapted the World3 system-dynamics model to a single island and ran population and welfare scenarios for Hawaii Island out to the year 2100. The paper appeared in Pacific Science, and a group at CERN invited me to Geneva to present it.

In 2014, a slow-moving lava flow from Kīlauea headed for the communities of Puna. I made maps and infographics to explain the flow to residents, and they spread much faster than I expected. The nonprofits organizing relief had no technical staff of their own. So I asked around, persuaded a vendor to donate a case-management system, and configured it for the coalition. They still use it. When Kīlauea erupted again in 2018, the Department of the Interior appointed me to its Strategic Sciences Group, 13 people asked to assess the consequences for the affected communities.

In between came two years as a data scientist at Hawaiian Airlines and five as a senior scientist at Oceanit Laboratories in Honolulu. At Oceanit I was principal investigator on Department of Defense and Department of Transportation projects, led a Department of Energy model of natural-disaster aftermath, and built edge-AI hardware that ran in the field. I also taught dozens of applied-AI workshops for Hawaii's students and teachers with the state Department of Education, straight through the COVID years.

Not everything from those years was work. I met my wife dancing Argentine tango in Honolulu, right before the pandemic. In 2023 I moved back to California, where I still am.

Press and media

My civic work during the 2014 and 2018 eruptions, and the Hawaii Island model, drew coverage in FiveThirtyEight, Hawaii Public Radio, Hawaii News Now, the Honolulu Star-Advertiser, the Hawaii Tribune-Herald, Xinhua News, KITV, Big Island Now, Hawaii 24/7, and University of Hawaii News.

Contact

Email me at mkimura1971@gmail.com, or find me on LinkedIn.