Job Description
Why Clarium
The healthcare industry overspends on its supply chain by over $25B each year — the result of fragmented data, inefficient workflows, and wasted supplies. Clarium is fixing that. Our AI-powered platform, Astra OS, gives hospitals end-to-end visibility into their supply chain operations, automating workflows and surfacing actionable insights so supply chain teams can focus on what matters most: patient care. We're trusted by some of the world's leading health systems, including Yale New Haven Health, Stanford, Geisinger, Cleveland Clinic, and Kaiser Permanente.
Founded in 2020, Clarium has raised $43M in total funding. Our Series A was led by Northzone, with participation from General Catalyst, AlleyCorp, Kaiser Permanente Ventures, Texas Medical Center Ventures, and 1984 Ventures.
The Opportunity
Clarium builds computer vision pipelines that extract structured data from clinical images under real-world conditions. This role owns the end-to-end pipeline: object detection, identification, reconciliation, and data extraction from images captured under variable lighting, camera angles, and workflow conditions with zero tolerance for errors.
You’ll design and build production-ready CV pipelines that combine state-of-the-art object detection models, multimodal LLM/LVM APIs, and barcode/label decoding to produce structured, auditable inventory data that clinical and supply chain workflows depend on. This has direct implications for patient safety, billing accuracy, and supply chain optimization.
In This Role You Will
Design, build, implement and optimize multi-stage CV pipelines spanning segmentation, object detection, multimodal LLM/LVM extraction, machine-readable code decoding, and multi-source reconciliation
Train or fine-tune detection models on custom medical supply datasets
Build and own dataset strategy - leverage augmentation and synthetic data generation to improve the training and testing datasets when data doesn’t exist.
Monitor and improve pipeline accuracy — instrument field-level metrics, diagnose failure modes, and systematically improve precision/recall through model iteration and preprocessing optimization
Design persistence schemas and audit data models that make every extraction independently reviewable
Maintain and extend the async Python backend services that surface pipeline results to downstream clinical workflows
What You’ll Bring
Required:
5+ years experience in computer vision and object detection
Hands-on experience training and fine-tuning detection models
Experience building OCR pipelines for label/packaging text extraction
Strong Python skills with experience in OpenCV, image preprocessing, and augmentation techniques
Production experience with multimodal LLM APIs for structured data extraction and validation
Backend engineering: FastAPI, Pydantic v2, PostgreSQL, async Python
Nice to Have:
OCR pipelines for label/packaging text extraction
Experience with barcode/QR/UDI decoding and preprocessing strategies that improve decode rates
MLOps experience: Docker, CI/CD, model versioning, A/B testing
Workflow orchestration tools (Temporal, Prefect, Airflow)
Healthcare or supply chain domain experience
Familiarity with medical device identification standards (UDI, GS1)
Skills & Tools You'll Use:
Need to Know: Python, PyTorch FastAPI · Pydantic v2 · PostgreSQL · Multimodal LLM APIs · Image preprocessing · Barcode / QR decoding, OCR pipelines
Nice to Know: Zero-shot object detection · Temporal · Prefect · Airflow
What You Get at Clarium
Incentive Stock Options proportionate to your salary
Fully remote — we're a distributed team across multiple time zones
Unlimited PTO
Top-tier health, vision, and dental benefits
401K
The opportunity to build on a strong foundational team with deep data and engineering roots at a stage where your work genuinely shapes the product