Radiology Vision-Language Pretraining Scientist

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

Job summary

Epsilon Labs, Inc. is seeking a Research Scientist to lead pretraining and scaling of vision-language models for radiology, across X-ray, CT, and MRI data. You will own the pretraining stage, design architectures, manage data mixtures, and drive model quality through evaluation.

You will work with PyTorch/JAX, large-scale distributed training, and contribute to publications and practical deployment in healthcare settings.

Qualifications

• 6+ years of academia/industry experience in vision-language modeling or related fields.
• Deep expertise in pretraining large vision-language models such as LLaVA, Flamingo, CogVLM, Qwen-VL, InternVL, or similar architectures.
• Strong foundation in modern VLM pretraining techniques including connector architectures, high-resolution handling, and embedding spaces.

Responsibilities

• Design, train, and scale vision-language foundation models for radiology applications, owning pretraining end-to-end.
• Develop architectures suited to medical imaging with high-resolution tiling and token budgets.
• Build and tune multimodal pretraining mixtures across captioning, VQA, grounding, and retrieval tasks.
• Develop fine-grained visual grounding during pretraining with bounding boxes or masks.
• Own pretraining evaluation (zero- and few-shot transfer, probing, and fine-tunability).
• Train joint vision-language embedding spaces using contrastive and generative objectives.

Skills

Vision-language modeling

Multimodal learning

Pretraining large models

Medical imaging

PyTorch

JAX

Distributed training

Long-context training

Grounding

Production-quality code

Tools

PyTorch

JAX

Job description

Epsilon Labs, Inc. is seeking a Research Scientist to lead pretraining and scaling of vision-language models for radiology, across X-ray, CT, and MRI data. You will own the pretraining stage, design architectures, manage data mixtures, and drive model quality through evaluation.

You will work with PyTorch/JAX, large-scale distributed training, and contribute to publications and practical deployment in healthcare settings.