Clinical Imaging Platform Engineer

About a2z Radiology AI

We build clinical AI that reads alongside radiologists. Our abdomen-pelvis CT triage device is FDA-cleared—the first commercial system to simultaneously triage seven urgent conditions on abdomen-pelvis CT in the U.S. We're backed by Khosla Ventures.

The Role

You'll own the infrastructure and interfaces that determine how good our diagnostic models can get: the quality of annotations and whether radiologists can see and trust model outputs. This is a product engineering role as much as infrastructure work.

Radiologist time is our most expensive input. When a reader needs four clicks instead of one, we lose annotation throughput, weaker models result, and critical findings get missed. The interface directly drives model quality.

Key Responsibilities

• Medical Imaging Viewer: Built on Cornerstone3D and VTK.js—manage volume rendering with stack rendering fallback, multiplanar reformats on demand, progressive loading from HTJ2K codestreams, and correctness features (radiological left/right, gantry tilt rulers, MONOCHROME1 inversion, codec handling)
• Annotation System: 3D mask storage, AI-assisted click-to-segment, classical segmentation tools (region growing, FWHM thresholding, multi-seed refinement), case-pool ledgers, phase state machines, ground-truth batches, peer-overlap evaluation, and agreement scorecards with cross-series resampling via NIfTI affines
• Annotation Schemas: Design per-lesion-category schemas with view, phase, and slice gating—made collaboratively with radiologists
• Model Output Visualization: Versioned sparse-RLE mask contracts keyed by SOP Instance UID; map model predictions through image position, direction cosines, and pixel spacing into anatomically correct world-space contours
• Quality Measurement: Build pipelines that measure annotation quality itself—race-condition-free case claiming, defined annotation phases, known ground-truth seeding, multi-reader batches, and per-lesion Dice scoring across different image geometries

What We're Looking For

You'll close a gap where one engineer currently holds this entire surface. Experience with medical imaging platforms, radiological workflows, 3D visualization (Cornerstone3D, VTK.js), and quality assurance pipelines is valuable. Familiarity with DICOM, NIfTI, HTJ2K, segmentation algorithms, and clinical annotation systems is a plus.