Writing
I write at Donuts with Chopsticks on Substack.
the gap series — benchmarks & evaluation
A Benchmark Score Is an Engine on a Standdraft
The power unit alone, and the same engine in a package — a wireframe F1 car where every part is a member of the clinical team. Engines don’t win championships; packages do.
One Chart, Five Questionsdraft
Healthcare AI benchmarks are saturated, and the evaluation that answers “does this help?” looks like product analytics — in private.
jump to: the race · the unasked question · the chart · the ticket · the wall
State of Healthcare AI Benchmarksdraft
Five things true at once: the record, the ceiling, the wall, and the instrument — the full landscape of healthcare AI evaluation in four parts.
Question Streamdraft
The precharting funnel rendered as live particles — swarm → the doctor → extraction and reasoning lanes → the usage log.
Model Evaluation as Product Analyticsdraft
Capability benchmarks are saturated; the evaluation that answers “does this help?” looks like product analytics, and it happens in private.
tutorials
Data Poetry: Intro to Biomedical Ontologies (Owlready2)draft
Ontologies as data poetry — reasoning over biomedical knowledge graphs in Python, landing on a rare-disease example.