Radiology Vision-Language AI Research Scientist

Epsilon Health; San Francisco (CA); On-site; USD 180,000 - 260,000; Full time

Job summary

Epsilon Health seeks a Research Scientist to lead the pretraining stage of radiology-focused vision-language models. You will own architecture design, data mixtures, and large-scale training across X-ray, CT, and MRI modalities, leveraging a large and diverse medical imaging dataset with clinical rigor.

The role emphasizes scaling, evaluation, and dissemination through publications and internal docs, collaborating with post-training teams to ensure production-quality results for clinical use.

Qualifications

• 6+ years of academia/industry experience in vision-language modeling, multimodal learning, or related fields
• Deep expertise in pretraining large vision-language models and multimodal fusion
• Experience with distributed training and producing production-ready code

Responsibilities

• Design, train, and scale vision-language foundation models for radiology applications, owning the pretraining stage end to end
• Develop VLM architectures suited to medical imaging with high-resolution tiling and token budgets
• Build multimodal pretraining mixtures across captioning, VQA, grounding and retrieval tasks
• Develop fine-grained visual grounding during pretraining and enable localization with boxes or masks
• Own pretraining evaluation and assess base model quality with transfer and fine-tuning analyses
• Train vision-language embeddings with contrastive and generative objectives across image-report pairs
• Contribute to dataset curation, architecture design, distributed training and checkpoint handoffs
• Stay current with vision-language modeling research and publish findings

Skills

Vision-language modeling

Multimodal learning

Pretraining large models

Research collaboration

Distributed training

Production-quality code

PyTorch

JAX

Medical imaging knowledge

Tools

PyTorch

JAX

Job description

Epsilon Health seeks a Research Scientist to lead the pretraining stage of radiology-focused vision-language models. You will own architecture design, data mixtures, and large-scale training across X-ray, CT, and MRI modalities, leveraging a large and diverse medical imaging dataset with clinical rigor.

The role emphasizes scaling, evaluation, and dissemination through publications and internal docs, collaborating with post-training teams to ensure production-quality results for clinical use.