Most people pick a lane, I keep looking for where they intersect.
Some days it's gene expression pipelines. Other days it's the exact border-radius that makes a button feel right. Occasionally both before the rising sun.
···················································································································································································································
Here is what I see:
The best interfaces borrow from biology. The best research borrows from taste.
And the best code is written by someone who reads more than documentation.
- ⟁ Wittgenstein-CLI — A modality harness for text-first LLMs, traceable end-to-end.
- ⧉ effectorHQ — AI capabilities are in the pre-TypeScript era. Effector adds a typed interface layer—tools can be compiled, verified, and composed across runtimes. All in one
effector.tomlsidecar. - ◒ ml-decision-boundary — Interactive ML visualization for seeing how SVM, Trees, KNN, and other models carve 2D feature space.
- ⟢ morpho — An ongoing multi-phase research pipeline for psychiatric epidemiology.
I don't specialize. I connect.




