Profile of Robert Hoekstra, the person behind awryLabs. A lawyer by training who builds machine-learning systems, a data and performance person in government, and a sociologist studying how communities hold together under disaster. awryLabs is the after-hours version of all of it: privacy-first games and tools. Links to GitHub at github.com/rhoekstr, LinkedIn at linkedin.com/in/rhoekstr, and email at coffee at awrylabs dot com.
Robert Hoekstra
Lawyer · data scientist · sociologist · maker
I find where systems fail people, then build small things to make them work better.
A lawyer by training who builds machine-learning systems, a data and performance
person in government by day, and a sociologist of how communities hold together, or
do not, when
a disaster suspends the usual rules. awryLabs is the after-hours version of all of it:
games and tools that are a little off on purpose.
The actual engine
Strip out the degrees and the job titles and you do not get a smaller version of me — you
get the actual engine. I treat understanding as the basic unit of a good life, and I have
spent a couple of decades turning that into a habit that will not switch off.
The pattern is not that I am curious about a lot of things. Plenty of people are. It is
that I go to the floor of whatever I pick up: I do not read across a field so much as read
its serious critics until I have finished them, and I do not take up a hobby so much as get
licensed in it. That produces an odd-looking résumé (economics, law, national security,
computer science, sociology) that reads like indecision until you notice it is one move
repeated: acquire exactly enough fluency in a discipline to force it into conversation with
the others. I find it genuinely intolerable that fields which obviously bear on each other
refuse to talk. So I keep building the bridges myself.
My actual research lives at one of those intersections: what happens to crime, and to a
community, when a disaster temporarily suspends the rules that normally govern behavior,
and why some places hold together while others fracture. It is a question you can only
answer by being a sociologist, a data person, and a policy person at once, which is most of
why I like it. And I aim the same lens inward. I treat myself as a system worth
understanding and tuning, not out of anxiety and not as a productivity flex, but because
clear thinking, paying real attention, and not being wrong in systematic ways are skills
you can actually practice. I would rather read the literature, build the framework, and
iterate than trust a vibe.
Off the clock
None of this would be sustainable if it were joyless, and it is not. There is a steady
diet of Wodehouse as the necessary counterweight, a lot of late-1990s UK electronic music,
an ongoing and slightly nerdy education in tea, board games, hiking in the western North
Carolina mountains where I live, and a family I am raising on the principle that you should
be allowed to read whatever you want, the more the better.
What I make, and why
awryLabs is what happens when that whole apparatus gets pointed at making things. The work
I value most makes a tangible difference for real people with a low risk of doing harm.
Extractive systems offend me more aesthetically than morally; I would simply rather build
the other kind. So the apps do not track you, do not nag you into a subscription, and try
to be a little strange on purpose rather than by accident. Five current projects, across
native apps, systems code, and the web, each with its own
backstory.
Swift · iOS / macOS / watchOS
Morass
A calm, turn-based territory game on living terrain. A small game about the complexity of power, with an honest accessibility story.
C++20 · Python
Gravel
A library for network-graph fragility analysis: roads, grids, telecom, any weighted graph. Born from disaster research and a single closed mountain road.
React
WOPR
Five turn-based strategy games on one menu, including a Global Thermonuclear War sim whose most common ending is the one where nobody plays.
React · TypeScript
Tyche
A psychedelic randomizer for small decisions: 27 lists, five animation styles, build-your-own, no telemetry. Built for a Disney-loving kid.
Vanilla JS · ML
Embedding Playground
An interactive demonstration that meaning is geometry, built to teach a workplace AI-literacy class that words matter.
Get in touch
For professional inquiries and work history, LinkedIn
is the place. For code, GitHub.
For anything else, email goes straight to my inbox, no form, no middleman.
✉ Email coffee@awrylabs.com