Member of Technical Staff, Software

Member of Technical Staff - AI Clinical Systems

San Francisco Bay Area (In-Person)

Metric is exclusively partnered with a frontier AI company building the infrastructure layer for clinical radiology. The company has raised $80M in backing from leading investors and is operating at the intersection of multimodal AI and real-world clinical deployment.

This team is building vision-language models that interpret full radiology studies in clinical context and generate draft reports used by radiologists in production.

The CTO previously built a vision-language model with 1M+ downloads on Hugging Face -reflecting real technical depth and adoption, not just research prototypes.

Their models are trained and validated on one of the largest real-world radiology datasets in existence, through partnership with a major radiology organization.

The Role:

This is not a feature role.

You will design and build the systems that convert:

• Imaging metadata
• Vision-language model outputs
• Radiologist voice dictation

into structured, production-grade clinical reports.

Expect to work on:

• Voice-to-report pipelines
• Structured document generation systems
• Low-latency AWS backend infrastructure
• ML inference integration at production scale

You will own services end-to-end - architecture, deployment, observability, reliability.

What They Look For:

They are not credential-driven. They look for exceptional builders.

• Experience shipping and operating production systems.
• Strong systems judgment in cloud environments (AWS preferred).
• Comfort designing API-driven, data-intensive architectures.
• Familiarity with distributed systems and reliability constraints.
• Experience working across structured and unstructured data.

Typical stack:

• AWS (ECS/EKS, S3, IAM, CloudWatch)
• Python backend services
• PostgreSQL

Strong Signals:

• Experience with document/report generation systems.
• Experience integrating ML inference into production workflows.
• Experience building systems with real-world correctness constraints.
• Healthcare or other regulated environment experience.

This is a team building foundational infrastructure for AI in diagnostic medicine - not a wrapper around models.