Most ML failures in healthcare aren't modeling failures. They're data governance decisions made too late.
We open source the tooling we build and ship the products that use it. Everything below is public — go look at the code.
Our Metal-based library for running neural networks on iOS. The reason on-device inference is a capability here and not a plan.
An open-source platform for running edge deep learning models on GPU, TPU, Jetson Nano and Coral.
An SDK unifying wearable health data across iOS and Android — the ingestion layer for anything you want to model on top of sensor data.
Optimizing clinical data structures for retrieval-augmented generation, so an LLM can answer over patient records without hallucinating the chart.
