Context
Fisheries scientists had four decades of trawl survey data showing that marine species were moving. Managers, journalists, and the public had almost no way to look at it. OceanAdapt was built to close that gap: a public database that lets anyone explore how the distribution of more than 650 marine species has changed over roughly forty years.
Decision
The hard problem was never storage. It was the distance between a scientific dataset and a question a person actually asks — where has this fish gone, and where is it going. I built the tool around that question rather than around the shape of the data, which meant committing to interactive maps, charts, and downloads as first-class features instead of as an export tacked onto a database.
Contribution
I was the engineering lead, credited in the project's own humans.txt. My work covers the parts that make scientific data usable: species search, map queries, distribution charts, full data downloads, future projections, seasonal views, and the comparison controls that let two species or two time periods be read side by side.
Result
OceanAdapt is public, free, and still running. Work originating in it was incorporated into the Fourth National Climate Assessment and later added as a U.S. Global Change Research Program indicator, and NOAA Fisheries and the U.S. Climate Resilience Toolkit both point the public to it. Those are other institutions choosing to build on the tool — the strongest evidence a public-interest project can have.
Learning
Early on I shipped a change that mixed several concerns in one commit, and noted at the time that it would be difficult to roll back. Nothing broke, but the lesson held: on a tool other institutions depend on, the ability to reverse a change quietly is worth more than the speed you gained by bundling it. Scoped pull requests, release branches, and reversible data operations became how I work, and the platform work I do now is built on that habit.