Dyania disclosed that Synapsis AI is built end to end on the NVIDIA AI Enterprise software stack: its proprietary clinical models are trained with the NVIDIA NeMo framework for training and reinforcement learning, and inference runs on NVIDIA TensorRT and TensorRT-LLM on the hospital's own NVIDIA GPU servers inside its private network. Confidential Computing, Dynamo and Multi-Instance GPU are named only as areas under evaluation and are not shipped.
Our readBuyers now have a named training and inference stack to assess, and a stated architecture in which patient records and inference stay on the health system's own hardware, which bears on PHI handling and infrastructure sizing.