ADHD flyouts
Small contextual thoughts can open beside a phrase without making the main thread leave the page.
I’m a founder at Poez Inc. and GoAIX, interested in the point where powerful technology becomes useful, understandable, and human.
I’m a founder and lead engineer at Poez Inc., where I build human-centered technology across AI systems, developer tooling, and game technology. I’m also a co-founder of GoAIX. Before these companies, I was a Principal Software Engineer at Microsoft, working across Windows Shell, AI, hardware, search, and edge experiences.
My career has moved through browser rendering, operating systems, graphics, game technology, machine learning, infrastructure, and open source. The domains change, but the throughline stays consistent: I like working close to the system while keeping sight of the person using it.
That systems-first interest reaches back to 2012, when I was researching artificial intelligence from the hardware upward - thinking about how neuron-like computation might be implemented efficiently in physical systems. I wrote about that line of inquiry in “Artificial Neurons: A Hardware Problem First”. I also published a code-backed study of 3D engines and 2D rasterization in an early graphics project.
This site is a public record of that work - projects I’ve built, systems I have helped ship, experiments I’m still thinking through, and writing about the decisions behind them.
Interface notes
This blog is also an instrument panel for how I think: associative, recursive, curious, and occasionally distracted by the interesting edge.
These are deliberate features, not glitches. Some are live, some are being built, and some are promises to the archive about the kinds of connections I want it to make visible.
Small contextual thoughts can open beside a phrase without making the main thread leave the page.
Ideas should connect to posts, projects, concepts, and career eras because a thought rarely lives alone.
A clean narrative and the raw notes behind it can both be honest ways to read the same work.
Technical ideas deserve more than one entrance: summary, engineer mode, and the deep rabbit hole.
Failed attempts, sketches, breakthroughs, and current versions are part of the artifact.
A manually curated view of the questions and systems currently taking up processing space.
Mechanisms matter, but so do the human consequences attached to every box and arrow.
The lab is where equations, rendering, pathfinding, and other ideas can move instead of merely being described.
Older writing and newer work should be allowed to recognize one another across time.
A playful signal for tangent density, equations, code, honesty, and how many rabbit holes are nearby.
Systems, games, spatial metaphors, and human interfaces are alternate doors into difficult ideas.
After one article, the next useful rabbit hole should come with a reason, not just a recommendation.