Adam Smith

At Abridge, I architect and evolve Linked Evidence, a clinical AI capability that lets clinicians trace AI-generated documentation to supporting conversation evidence. I also contribute backend services and data workflows for Ambient Clinical Documentation.

I build evidence-grounded workflows for ambient clinical intelligence, investigate failure modes such as unsupported claims and missing context, and establish clinician-in-the-loop evaluation processes for production healthcare AI.

Previously at Palantir, I delivered customer-facing Foundry solutions, using Ontology models and enterprise data integrations to turn fragmented operational data into production workflow applications. At YouTube, I developed backend components for Procella, supporting high-volume SQL analytics and distributed data-processing workloads.

I'm motivated by AI products that reduce clinicians' administrative burden while making medical documentation more transparent, reliable, and actionable.