Research Scientist - Vision Foundation Models

Epsilon Health; San Francisco (CA); On-site; USD 180,000 - 240,000; Full time

Job summary

Epsilon Health in San Francisco seeks a Research Scientist with deep expertise in vision foundation models to advance AI for medical imaging across X-ray, CT, and MRI.

You will pretrain and scale encoders, extend 2D to volumetric data, evaluate rigorously, and contribute to production deployment while publishing results.

Qualifications

• 6+ years of academia/industry experience in computer vision/machine learning.
• Deep expertise in training vision encoder models at scale (e.g. ViT, ConvNeXt). Strong foundation in self-supervised pretraining, including contrastive, masked image modeling, self-distillation, and JEPA-style objectives.
• Experience training on volumetric or spatiotemporal data (video, 3D medical imaging).
• Track record of implementing complex models from research papers and adapting them to new domains.
• Proficiency in PyTorch or JAX, with experience training models on multi-GPU/distributed systems.
• Hands-on experience with medical imaging applications, particularly radiology (X-ray, CT, MRI).
• Strong software engineering skills and ability to write production-quality code.

Responsibilities

• Design, train, and scale vision foundation models for radiology applications across X-ray, CT, and MRI modalities, implementing self-supervised, contrastive, masked image modeling, and joint-embedding predictive (JEPA) frameworks.
• Extend 2D pretraining recipes to volumetric CT and MR data, addressing long sequence lengths, anisotropic spacing, and multi-sequence studies.
• Evaluate model performance rigorously across academic benchmarks, internal offline datasets, and live production data.
• Contribute hands-on to all stages of model development including dataset curation, architecture design, distributed training, and production deployment.
• Stay current with cutting-edge research in computer vision and medical imaging AI.
• Drive research and technical excellence through conference publications and technical blog posts, establishing best practices for training robust medical imaging models at scale.

Skills

Experience in CV/ML

Vision encoders at scale

Self-supervised pretraining

JEPA objectives

Volumetric data

Model implementation

PyTorch/JAX

Medical imaging

Production-grade coding

Job description

About Us

We're tackling one of healthcare's most critical challenges in medical imaging and diagnostics. Our company operates at the intersection of cutting-edge AI and clinical practice, building technology that directly impacts patient outcomes. We've assembled one of the industry's most comprehensive and diverse medical imaging datasets and have a proven product-market fit with a substantial customer pipeline already in place.

About Us

We're tackling one of healthcare's most critical challenges in medical imaging and diagnostics. Our company operates at the intersection of cutting-edge AI and clinical practice, building technology that directly impacts patient outcomes. We've assembled one of the industry's most comprehensive and diverse medical imaging datasets and have a proven product-market fit with a substantial customer pipeline already in place.

Role Overview

We're seeking a Research Scientist with deep expertise in vision foundation models to join our ML Research team. You'll be at the forefront of developing and deploying state-of-the-art vision models for medical imaging applications. This role focuses on pretraining and scaling vision encoders for radiology diagnosis across X-ray, CT, and MRI, with a growing emphasis on 3D volumetric modeling. You'll work with one of the largest and most diverse medical imaging datasets in the industry, pushing the boundaries of what's possible in AI-assisted diagnosis while maintaining the rigor required for clinical deployment.

Key Responsibilities

• Design, train, and scale vision foundation models for radiology applications across X-ray, CT, and MRI modalities, implementing self-supervised, contrastive, masked image modeling, and joint-embedding predictive (JEPA) frameworks.
• Extend 2D pretraining recipes to volumetric CT and MR data, addressing long sequence lengths, anisotropic spacing, and multi-sequence studies.
• Evaluate model performance rigorously across academic benchmarks, internal offline datasets, and live production data.
• Contribute hands-on to all stages of model development including dataset curation, architecture design, distributed training, and production deployment.
• Stay current with cutting-edge research in computer vision and medical imaging AI.
• Drive research and technical excellence through conference publications and technical blog posts, establishing best practices for training robust medical imaging models at scale.

Qualifications

• 6+ years of academia/industry experience in computer vision/machine learning
• Deep expertise in training vision encoder models at scale (e.g. ViT, ConvNeXt). Strong foundation in self-supervised pretraining, including contrastive, masked image modeling, self-distillation, and JEPA-style objectives.
• Experience training on volumetric or spatiotemporal data (video, 3D medical imaging)
• Track record of implementing complex models from research papers and adapting them to new domains
• Proficiency in PyTorch or JAX, with experience training models on multi-GPU/distributed systems
• Hands-on experience with medical imaging applications, particularly radiology (X-ray, CT, MRI)
• Strong software engineering skills and ability to write production-quality code

Preferred Qualifications

• Publications at top-tier conferences (CVPR, ICCV/ECCV, NeurIPS, ICLR, MICCAI)
• Experience with 3D medical image processing and retrieval tasks
• Familiarity with CT and MR acquisition (windowing, multi-sequence protocols, voxel spacing)
• Experience with long-context training techniques (sequence parallelism, efficient attention)
• Knowledge of vision-language models and multimodal learning
• Experience with model interpretability and explainability methods
• Understanding of clinical evaluation metrics, clinical workflows, and healthcare data (DICOM, HL7, etc.)