Teams at a2z Radiology AI

Develop end-to-end foundation models that read entire CT studies and generate radiology-style reports. Build vision-language and 3D models, run large-scale distributed training and rigorous experiments, and deliver models that progress through FDA clearance to hospital deployment. Collaborate with clinicians and engineers to ensure safety, reproducibility, and clinical impact.

Build and own the clinical imaging viewer and annotation platform used by radiologists: implement volume and stack rendering, progressive HTJ2K loading, DICOM geometry correctness, 3D mask storage, AI-assisted segmentation tools, annotation state machines and agreement scoring, schema gating, versioned model mask contracts, and secure FastAPI backend with HIPAA-compliant access controls and cohort isolation.

Own end-to-end clinical data ingest and result delivery for a radiology AI product: on‑prem C‑STORE ingest, cloud DICOMweb STOW-RS handling, tenancy/privacy at the boundary, outbound HL7/JSON delivery, per-study tracing and health checks, CI for gateway repos, and precise security-review documentation. Troubleshoot protocol, networking, and memory/streaming issues using packet captures and profiling.