Maziyar Panahi

At ISC-PIF, Panahi’s work includes high-performance computing, big-data platforms and infrastructure management. He previously led the team behind John Snow Labs’ Spark NLP, developing components for distributed NLP and model deployment in the JVM and Apache Spark ecosystem. His personal account places that work from 2019 to 2025 and his Arcee post-training work from 2024 onward.[1][3][4]

OpenMed’s local runtime extracts clinical entities and removes identifying information after its model artifacts are available. It exposes Python and command-line workflows alongside supported mobile and service interfaces. The repository distinguishes that local path from model downloads, remote adapters and other configured integrations that can use a network.[2]

Make validation a deployment responsibility

The project documents limits rather than treating use of the package as proof of privacy or clinical fitness. Models and datasets have their own terms, and deployments need review of their behavior. Its synthetic brief demonstration explicitly uses test doubles and is not clinical validation. These distinctions connect an inspectable implementation with the work needed to assess it in a particular setting.[2]

Keep medical models open to inspection

Panahi argues that medical AI should be open, auditable and deployable within a hospital’s own environment. He connects that position with releasing transparent models through OpenMed and with the infrastructure needed to operate them. His account treats model development and deployment as connected work, spanning fine-tuning, data platforms and the systems that run them.[3]